refactor(import): 导入服务核心逻辑改造

- 预览阶段增加 Excel 内部重复检测 + 空规格去重 + 字段长度校验
- 执行阶段支持 skip 模式（跳过重复行继续导入）+ update 模式（覆盖更新已有记录）
- update 模式空值保护：Excel 空单元格不覆盖已有数据
- DB commit 异常防御：字段超长等异常优雅返回而非 500
- 去重规则改为仅依据 spec_model（货号唯一键）
- BOM 导入物料查找同步改为 spec-only

inventory-backend/app/services/import_service.py

@@ -116,8 +116,14 @@ def generate_template(import_type: str) -> io.BytesIO:
     return output
 
 
-def preview_material(file_stream) -> list:
+def preview_material(file_stream, mode: str = 'skip') -> list:
-    """Dry-Run 预览基础信息导入"""
+    """Dry-Run 预览基础信息导入（含 Excel 内部重复检测 + 空规格去重）
+
+    Args:
+        file_stream: Excel 文件流
+        mode: 'skip' — DB 重复视为错误（默认）
+              'update' — DB 重复标记为 'update' 状态（将被覆盖更新，不视为错误）
+    """
     import pandas as pd
 
     df = pd.read_excel(file_stream, header=0)
@@ -138,15 +144,19 @@ def preview_material(file_stream) -> list:
     }
     df.rename(columns=col_map, inplace=True)
 
-    # 批量预查已存在的 name+spec 组合（用于重复检测）
+    # 批量预查已存在的 spec_model（作为唯一货号，用于重复检测）
-    existing = db.session.query(MaterialBase.name, MaterialBase.spec_model).all()
+    existing = db.session.query(MaterialBase.spec_model).all()
-    existing_set = {(r.name, r.spec_model or '') for r in existing}
+    existing_set = {r.spec_model for r in existing if r.spec_model}
+
+    # Excel 文件内部去重映射（spec → 首次出现行号）
+    excel_seen_map = {}
 
     results = []
     for idx, row in df.iterrows():
         row_num = idx + 2  # Excel 行号
         errors = []
         name = str(row.get('name', '')).strip() if pd.notna(row.get('name')) else ''
+        # spec_model 作为唯一货号，空值严格拦截
         spec = str(row.get('spec', '')).strip() if pd.notna(row.get('spec')) else ''
         category = str(row.get('category', '')).strip() if pd.notna(row.get('category')) else ''
         mat_type = str(row.get('type', '')).strip() if pd.notna(row.get('type')) else ''
@@ -155,7 +165,7 @@ def preview_material(file_stream) -> list:
         if not name:
             errors.append("名称不能为空")
         if not spec:
-            errors.append("规格型号不能为空")
+            errors.append("缺少必填项: 规格型号 (作为唯一货号不能为空)")
         if not category:
             errors.append("类别不能为空")
         if not mat_type:
@@ -163,8 +173,40 @@ def preview_material(file_stream) -> list:
         if not unit:
             errors.append("计量单位不能为空")
 
-        if name and spec and (name, spec) in existing_set:
+        # ★ 新增：字段长度校验（预防 DB commit 时因字段超长回滚）
-            errors.append(f"已存在相同名称和规格的数据: {name} / {spec}")
+        MAX_LENS = {
+            '名称': (name, 255), '规格型号': (spec, 255),
+            '所属公司': (str(row.get('company_name', '')) if pd.notna(row.get('company_name')) else '', 255),
+            '专业名称': (str(row.get('common_name', '')) if pd.notna(row.get('common_name')) else '', 255),
+            '类别': (category, 100), '类型': (mat_type, 100),
+            '计量单位': (unit, 50),
+        }
+        for field_label, (val, max_len) in MAX_LENS.items():
+            if len(val) > max_len:
+                errors.append(f"{field_label}超出数据库限制({len(val)}>{max_len}字符)")
+
+        # ★ 去重检测（仅依据 spec_model 作为唯一货号）
+        if spec:
+            is_db_dup = spec in existing_set
+
+            if is_db_dup:
+                if mode == 'update':
+                    # 覆盖更新模式：DB 重复不视为错误
+                    if spec in excel_seen_map:
+                        first_row = excel_seen_map[spec]
+                        errors.append(f"与表格内第 {first_row} 行的规格型号重复: {spec}")
+                    else:
+                        excel_seen_map[spec] = row_num
+                else:
+                    # 跳过模式：DB 重复是错误
+                    errors.append(f"数据库已存在相同规格型号 (货号) 的数据: {spec}")
+            else:
+                # 不在 DB 中 → 检查是否与 Excel 内前面行重复
+                if spec in excel_seen_map:
+                    first_row = excel_seen_map[spec]
+                    errors.append(f"与表格内第 {first_row} 行的规格型号重复: {spec}")
+                else:
+                    excel_seen_map[spec] = row_num
 
         company = str(row.get('company_name', '')).strip() if pd.notna(row.get('company_name')) else ''
         common = str(row.get('common_name', '')).strip() if pd.notna(row.get('common_name')) else ''
@@ -173,10 +215,18 @@ def preview_material(file_stream) -> list:
         is_enabled = parse_boolean(row.get('is_enabled', True))
         is_inspection = parse_boolean(row.get('is_inspection_required', False))
 
+        # 决定行状态：update 模式下 DB 重复但不含其他错误的行 → 'update'
+        if mode == 'update' and not errors and spec and spec in existing_set:
+            row_status = 'update'
+        elif errors:
+            row_status = 'error'
+        else:
+            row_status = 'success'
+
         results.append({
             'row': row_num,
-            'status': 'error' if errors else 'success',
+            'status': row_status,
-            'error_msg': '; '.join(errors) if errors else '',
+            'error_msg': '; '.join(errors) if errors else ('将被覆盖更新' if row_status == 'update' else ''),
             'data': {
                 'company_name': company,
                 'name': name,
@@ -221,14 +271,14 @@ def preview_bom(file_stream) -> list:
     }
     df.rename(columns=col_map, inplace=True)
 
-    # 批量预查所有 MaterialBase
+    # 批量预查所有 MaterialBase（仅依据 spec_model 查找）
     all_materials = db.session.query(
         MaterialBase.id, MaterialBase.name, MaterialBase.spec_model
     ).all()
     mat_lookup = {}
     for m in all_materials:
-        key = (m.name, m.spec_model or '')
+        if m.spec_model:
-        mat_lookup[key] = m.id
+            mat_lookup[m.spec_model] = m.id
 
     # 预查所有已存在的 (bom_no, version, parent_id, child_id) 组合
     all_bom_pairs = db.session.query(
@@ -259,15 +309,13 @@ def preview_bom(file_stream) -> list:
         if pd.isna(dosage) or float(dosage) <= 0:
             errors.append("用量必须大于0")
 
-        parent_key = (parent_name, parent_spec)
+        parent_id = mat_lookup.get(parent_spec) if parent_spec else None
-        parent_id = mat_lookup.get(parent_key) if parent_name else None
         if parent_name and parent_id is None:
-            errors.append(f"父件不存在: {parent_name} / {parent_spec}")
+            errors.append(f"父件不存在 (规格型号): {parent_spec}")
 
-        child_key = (child_name, child_spec)
+        child_id = mat_lookup.get(child_spec) if child_spec else None
-        child_id = mat_lookup.get(child_key) if child_name else None
         if child_name and child_id is None:
-            errors.append(f"子件不存在: {child_name} / {child_spec}")
+            errors.append(f"子件不存在 (规格型号): {child_spec}")
 
         if parent_id and child_id and parent_id == child_id:
             errors.append("父件与子件不能是同一物料")
@@ -298,20 +346,74 @@ def preview_bom(file_stream) -> list:
     return results
 
 
-def execute_material_import(rows: list) -> dict:
+def execute_material_import(rows: list, mode: str = 'skip') -> dict:
-    """执行基础信息批量导入（事务保护）"""
+    """执行基础信息批量导入
+
+    Args:
+        rows: 待导入行数据
+        mode: 'skip' - 跳过重复行，继续导入其余数据（默认）
+              'update' - 匹配到重复时覆盖更新已有记录
+    """
+    import pandas as pd
+
+    def _has_value(val) -> bool:
+        """判断值是否有效（Excel 单元格非空、非 NaN）"""
+        if val is None:
+            return False
+        try:
+            if pd.isna(val):
+                return False
+        except Exception:
+            pass
+        s = str(val).strip()
+        return s != '' and s != 'nan' and s != 'None'
+
     inserted = 0
+    updated = 0
+    skipped = 0
     errors = []
 
     try:
         for i, item in enumerate(rows):
             d = item.get('data', item)
+            spec = d.get('spec_model', '')
             try:
-                existing = MaterialBase.query.filter_by(
+                # ★ 仅依据 spec_model（唯一货号）查找已有记录
-                    name=d['name'], spec_model=d['spec_model']
+                existing = MaterialBase.query.filter_by(spec_model=spec).first() if spec else None
-                ).first()
+
                 if existing:
-                    errors.append({'row': i + 2, 'msg': f"已存在: {d['name']} / {d['spec_model']}"})
+                    if mode == 'update':
+                        # ★ 覆盖更新模式：用 Excel 数据更新已有记录（含名称）
+                        # 仅当 Excel 单元格有实际值时覆盖，空值不覆盖已有数据
+                        if _has_value(d.get('name')):
+                            existing.name = str(d['name']).strip()
+                        if _has_value(d.get('company_name')):
+                            existing.company_name = str(d['company_name']).strip()
+                        if _has_value(d.get('common_name')):
+                            existing.common_name = str(d['common_name']).strip()
+                        if _has_value(d.get('category')):
+                            existing.category = str(d['category']).strip()
+                        if _has_value(d.get('material_type')):
+                            existing.material_type = str(d['material_type']).strip()
+                        if _has_value(d.get('spec_model')):
+                            existing.spec_model = str(d['spec_model']).strip()
+                        if _has_value(d.get('unit')):
+                            existing.unit = str(d['unit']).strip()
+                        if d.get('reference_price') is not None and _has_value(d.get('reference_price')):
+                            try:
+                                existing.reference_price = float(d['reference_price'])
+                            except Exception:
+                                pass
+                        if _has_value(d.get('is_inspection_required')):
+                            existing.is_inspection_required = parse_boolean(d.get('is_inspection_required'))
+                        if _has_value(d.get('is_enabled')):
+                            existing.is_enabled = parse_boolean(d.get('is_enabled'))
+                        updated += 1
+                        continue
+                    else:
+                        # ★ 跳过模式（默认）：记录跳过，继续导入其余行
+                        skipped += 1
+                        errors.append({'row': i + 2, 'msg': f"重复数据已跳过 (货号): {spec}"})
                         continue
 
                 mat = MaterialBase(
@@ -331,18 +433,43 @@ def execute_material_import(rows: list) -> dict:
             except Exception as e:
                 errors.append({'row': i + 2, 'msg': str(e)})
 
-        if errors:
+        # ★ 改造：不再"一刀切"回滚，只提交成功的行
-            db.session.rollback()
+        # 防御：捕获 DB 级异常（字段超长、约束冲突等），优雅返回而非 500
-            return {'success': False, 'inserted': 0, 'errors': errors,
+        try:
-                    'msg': f'导入失败，{len(errors)} 条数据校验未通过'}
-
             db.session.commit()
-        return {'success': True, 'inserted': inserted, 'errors': [],
+        except Exception as commit_err:
-                'msg': f'成功导入 {inserted} 条基础信息'}
+            db.session.rollback()
+            return {
+                'success': False,
+                'inserted': 0,
+                'updated': 0,
+                'skipped': 0,
+                'errors': [{'row': 0, 'msg': f'数据库写入失败: {str(commit_err)}'}],
+                'msg': f'导入失败，数据库写入异常（可能字段超长或约束冲突）: {str(commit_err)}',
+            }
+
+        # 构建结果消息
+        parts = []
+        if inserted > 0:
+            parts.append(f"新增 {inserted} 条")
+        if updated > 0:
+            parts.append(f"更新 {updated} 条")
+        if skipped > 0:
+            parts.append(f"跳过 {skipped} 条重复")
+        msg = "成功: " + "，".join(parts) if parts else "没有数据被导入"
+
+        return {
+            'success': True,
+            'inserted': inserted,
+            'updated': updated,
+            'skipped': skipped,
+            'errors': errors,
+            'msg': msg,
+        }
 
     except Exception as e:
         db.session.rollback()
-        return {'success': False, 'inserted': 0,
+        return {'success': False, 'inserted': 0, 'updated': 0, 'skipped': 0,
                 'errors': [{'row': 0, 'msg': str(e)}], 'msg': f'导入异常: {str(e)}'}
 
 
@@ -354,7 +481,8 @@ def execute_bom_import(rows: list) -> dict:
     ).all()
     mat_lookup = {}
     for m in all_materials:
-        mat_lookup[(m.name, m.spec_model or '')] = m.id
+        if m.spec_model:
+            mat_lookup[m.spec_model] = m.id
 
     inserted = 0
     errors = []
@@ -369,11 +497,11 @@ def execute_bom_import(rows: list) -> dict:
                 version = d.get('version', 'V1.0')
 
                 if not parent_id:
-                    pk = (d.get('parent_name', ''), d.get('parent_spec', ''))
+                    parent_spec = d.get('parent_spec', '')
-                    parent_id = mat_lookup.get(pk)
+                    parent_id = mat_lookup.get(parent_spec) if parent_spec else None
                 if not child_id:
-                    ck = (d.get('child_name', ''), d.get('child_spec', ''))
+                    child_spec = d.get('child_spec', '')
-                    child_id = mat_lookup.get(ck)
+                    child_id = mat_lookup.get(child_spec) if child_spec else None
 
                 if not parent_id or not child_id:
                     errors.append({'row': i + 2, 'msg': '父件或子件在系统中不存在'})