SAP ABAP 内表全面详解

一、内表基本概念

1.1 什么是内表？

内表是ABAP程序运行时在内存中创建的临时表，用于：

• 存储和处理多行数据
• 作为程序间数据传递的容器
• 与数据库表进行批量数据交换
• 进行复杂的数据计算和处理

1.2 内表的物理结构

内存中的内表结构： ┌─────────────────────────────────┐ │ 内表控制块 (Table Control Block) │ ├─────────────────────────────────┤ │ 表头信息 (Header Information) │ │ - 行数 │ │ - 行宽度 │ │ - 表类型 │ │ - 键信息 │ ├─────────────────────────────────┤ │ 数据行指针数组 │ │ ┌─┬─┬─┬─┬─┬─┐ │ │ │0│1│2│3│4│5│ → 指向实际数据行 │ │ └─┴─┴─┴─┴─┴─┘ │ ├─────────────────────────────────┤ │ 实际数据存储区 │ │ ┌─────────────────┐ │ │ │ 行1数据 │ │ │ ├─────────────────┤ │ │ │ 行2数据 │ │ │ ├─────────────────┤ │ │ │ ... │ │ │ └─────────────────┘ │ └─────────────────────────────────┘

1.3 内表与数据库表的区别

二、内表类型深度解析

2.1 标准表（STANDARD TABLE）

" 标准表的完整声明语法 DATA: gt_std TYPE STANDARD TABLE OF sflight WITH NON-UNIQUE KEY carrid connid WITH HEADER LINE INITIAL SIZE 100 WITH UNIQUE SORTED KEY sec_key COMPONENTS fldate WITH NON-UNIQUE SORTED KEY sec_key2 COMPONENTS price currency.

特点：

• 默认表类型
• 支持索引访问（INDEX）和键访问（WITH KEY）
• 使用线性查找算法
• 插入行效率高，查找效率低（O(n)）

2.2 排序表（SORTED TABLE）

" 排序表的声明 DATA: gt_sorted TYPE SORTED TABLE OF sflight WITH UNIQUE KEY carrid connid fldate WITH NON-UNIQUE SORTED KEY price_key COMPONENTS price INITIAL SIZE 50.

特点：

• 数据始终按键排序存储
• 使用二分查找算法（O(log n)）
• 插入/删除需要维护排序
• 可定义唯一或非唯一主键

2.3 哈希表（HASHED TABLE）

" 哈希表的声明 DATA: gt_hashed TYPE HASHED TABLE OF sflight WITH UNIQUE KEY mandt carrid connid fldate WITH NON-UNIQUE HASHED KEY carrier_key COMPONENTS carrid INITIAL SIZE 200.

特点：

• 键访问最快（O(1)）
• 不支持索引访问
• 内存消耗较大
• 键必须是唯一的

2.4 索引表（INDEX TABLE）- 已弃用但需了解

" 传统索引表声明 DATA: gt_index TYPE INDEX TABLE OF sflight WITH UNIQUE KEY carrid OCCURS 0.

• OCCURS子句已弃用
• 用STANDARD TABLE替代

三、内表声明方式大全

3.1 基于ABAP字典对象

" 1. 基于透明表 DATA: gt_sflight TYPE TABLE OF sflight. DATA: gt_sflight TYPE STANDARD TABLE OF sflight. " 2. 基于结构 DATA: gt_mara TYPE TABLE OF mara. DATA: gt_ekpo TYPE TABLE OF ekpo. " 3. 基于类型池（已弃用，了解即可） TYPE-POOLS: slis. DATA: gt_fieldcat TYPE slis_t_fieldcat_alv.

3.2 基于本地类型

" 1. 先定义类型，再声明变量 TYPES: BEGIN OF ty_employee, pernr TYPE persno, name TYPE string, dept TYPE orgeh, salary TYPE betrg, END OF ty_employee. TYPES: tt_employees TYPE STANDARD TABLE OF ty_employee WITH NON-UNIQUE KEY pernr. DATA: gt_employees TYPE tt_employees. " 2. 直接声明 DATA: gt_employees TYPE TABLE OF ty_employee.

3.3 使用TYPE REF TO声明内表引用

" 内表对象引用 DATA: go_table TYPE REF TO data. DATA: gt_data TYPE REF TO data. FIELD-SYMBOLS: <fs_table> TYPE ANY TABLE. " 动态创建内表 CREATE DATA go_table TYPE TABLE OF (dyn_table_name). ASSIGN go_table->* TO <fs_table>.

3.4 带表头行 vs 不带表头行

" 带表头行（旧式，不推荐） DATA: gt_itab TYPE TABLE OF sflight WITH HEADER LINE. gt_itab-carrid = 'LH'. " 访问表头行 APPEND gt_itab. " 将表头行添加到表体 " 不带表头行（推荐） DATA: gt_itab TYPE TABLE OF sflight. DATA: gs_wa TYPE sflight. gs_wa-carrid = 'LH'. APPEND gs_wa TO gt_itab.

四、内表操作全解

4.1 填充操作

APPEND

" 基本APPEND APPEND gs_line TO gt_table. " APPEND INITIAL LINE - 添加初始行 APPEND INITIAL LINE TO gt_table ASSIGNING FIELD-SYMBOL(<fs>). <fs>-field = 'value'. " APPEND LINES OF - 批量添加 DATA: gt_source TYPE TABLE OF sflight, gt_target TYPE TABLE OF sflight. SELECT * FROM sflight INTO TABLE gt_source. APPEND LINES OF gt_source TO gt_target. " APPEND 使用排序键（仅排序表） DATA: gt_sorted TYPE SORTED TABLE OF sflight WITH UNIQUE KEY carrid connid. gs_line-carrid = 'AA'. gs_line-connid = '0017'. APPEND gs_line TO gt_sorted. " 自动插入到正确位置

INSERT

" INSERT INTO TABLE - 智能插入 INSERT gs_line INTO TABLE gt_table. " INSERT INTO INDEX - 指定位置插入 INSERT gs_line INTO gt_table INDEX 5. " INSERT LINES OF - 批量插入 INSERT LINES OF gt_source FROM 1 TO 10 INTO gt_target INDEX 1. " 使用字段符号插入 INSERT VALUE #( carrid = 'LH' connid = '0400' ) INTO TABLE gt_table.

COLLECT

TYPES: BEGIN OF ty_sum, matnr TYPE matnr, menge TYPE menge_d, meins TYPE meins, END OF ty_sum. DATA: gt_sum TYPE TABLE OF ty_sum WITH NON-UNIQUE KEY matnr. gs_sum-matnr = 'MAT001'. gs_sum-menge = 10. gs_sum-meins = 'PC'. COLLECT gs_sum INTO gt_sum. gs_sum-matnr = 'MAT001'. gs_sum-menge = 5. COLLECT gs_sum INTO gt_sum. " MAT001的menge变为15

4.2 读取操作

READ TABLE的各种形式

" 1. 通过索引读取 READ TABLE gt_table INTO gs_line INDEX 5. " 2. 通过键读取 READ TABLE gt_table INTO gs_line WITH KEY carrid = 'LH'. READ TABLE gt_table INTO gs_line WITH KEY carrid = 'LH' connid = '0400'. " 3. 通过表键读取（排序表和哈希表） READ TABLE gt_sorted INTO gs_line WITH TABLE KEY carrid = 'LH' connid = '0400'. " 4. 使用BINARY SEARCH（必须排序） SORT gt_table BY carrid connid. READ TABLE gt_table INTO gs_line WITH KEY carrid = 'LH' connid = '0400' BINARY SEARCH. " 5. 读取到字段符号（高效） READ TABLE gt_table ASSIGNING FIELD-SYMBOL(<fs_line>) INDEX 1. IF sy-subrc = 0. <fs_line>-field = 'new value'. ENDIF. " 6. 读取到数据引用 DATA: lr_line TYPE REF TO data. READ TABLE gt_table REFERENCE INTO lr_line INDEX 1. IF sy-subrc = 0. ASSIGN lr_line->* TO FIELD-SYMBOL(<fs>). ENDIF. " 7. 读取并传输字段 READ TABLE gt_table INDEX 1 TRANSPORTING carrid connid. " 8. 读取时忽略缓存 READ TABLE gt_table INTO gs_line WITH KEY carrid = 'LH' BYPASSING BUFFER. " 忽略表缓冲

内表表达式读取（7.4+）

" 1. 直接读取 DATA(ls_line) = gt_table[ 1 ]. " 索引读取 DATA(ls_line2) = gt_table[ carrid = 'LH' ]. " 键读取 " 2. 安全读取（避免异常） DATA(ls_line3) = VALUE #( gt_table[ 99 ] OPTIONAL ). " 索引 DATA(ls_line4) = VALUE #( gt_table[ carrid = 'XXX' ] OPTIONAL ). " 键 " 3. 使用辅助键读取 DATA(ls_line5) = gt_table[ KEY sec_key price = 1000 ]. " 4. 读取并修改 gt_table[ 1 ]-field = 'new value'.

4.3 修改操作

MODIFY

" 1. 通过索引修改 MODIFY gt_table FROM gs_line INDEX 3. " 2. 通过表键修改 MODIFY TABLE gt_table FROM gs_line. " 使用主键 " 3. 通过键修改（修改所有匹配行！） MODIFY gt_table FROM gs_line TRANSPORTING price WHERE carrid = 'LH'. " 危险：修改所有LH的行 " 4. 在循环中修改 LOOP AT gt_table ASSIGNING FIELD-SYMBOL(<fs_line>) WHERE carrid = 'LH'. <fs_line>-price = <fs_line>-price * 1.1. ENDLOOP.

UPDATE

" UPDATE仅用于数据库表，内表用MODIFY UPDATE sflight FROM TABLE gt_sflight. " 更新数据库表

4.4 删除操作

DELETE

" 1. 通过索引删除 DELETE gt_table INDEX 5. " 2. 通过表键删除 DELETE TABLE gt_table WITH TABLE KEY carrid = 'LH' connid = '0400'. " 3. 通过键删除 DELETE gt_table WHERE carrid = 'LH'. " 删除所有匹配行 " 4. 删除相邻重复行 SORT gt_table BY carrid connid fldate. DELETE ADJACENT DUPLICATES FROM gt_table COMPARING carrid connid. " 5. 删除重复行但保留第一行 DELETE ADJACENT DUPLICATES FROM gt_table COMPARING ALL FIELDS. " 6. 批量删除 DELETE gt_table FROM 10 TO 20. " 删除索引10-20的行

4.5 循环遍历

LOOP AT

" 1. 基本循环 LOOP AT gt_table INTO gs_line. " 处理数据 ENDLOOP. " 2. 使用字段符号（推荐） LOOP AT gt_table ASSIGNING FIELD-SYMBOL(<fs_line>). <fs_line>-field = 'value'. ENDLOOP. " 3. 使用WHERE条件 LOOP AT gt_table INTO gs_line WHERE carrid = 'LH'. WRITE: / gs_line-carrid, gs_line-connid. ENDLOOP. " 4. 使用FROM/TO限定范围 LOOP AT gt_table INTO gs_line FROM 5 TO 10. ENDLOOP. " 5. 循环并删除（需使用索引） LOOP AT gt_table ASSIGNING FIELD-SYMBOL(<fs_line>). IF <fs_line>-carrid = 'XX'. DELETE gt_table INDEX sy-tabix. ENDIF. ENDLOOP. " 6. 循环使用TRANSPORTING LOOP AT gt_table INTO gs_line TRANSPORTING carrid connid. ENDLOOP.

内表表达式循环（7.4+）

" 1. 使用FOR表达式 DATA(lt_filtered) = VALUE tt_sflight( FOR wa IN gt_sflight WHERE ( carrid = 'LH' ) ( wa ) ). " 2. 在构造函数中使用循环 DATA(lt_names) = VALUE string_table( FOR wa IN gt_sflight ( |{ wa-carrid }-{ wa-connid }| ) ).

4.6 排序与分组

SORT

" 1. 基本排序 SORT gt_table BY carrid ASCENDING connid DESCENDING. " 2. 按数字字段排序 SORT gt_table BY price AS TEXT. " 文本排序 SORT gt_table BY price AS NUMBER. " 数字排序 " 3. 稳定排序 SORT gt_table STABLE BY carrid. " 保持相等元素的相对顺序 " 4. 自定义排序顺序 TYPES: BEGIN OF ty_sort_order, field TYPE string, order TYPE c LENGTH 1, END OF ty_sort_order. DATA: gt_sort TYPE TABLE OF ty_sort_order. APPEND VALUE #( field = 'CARRID' order = 'A' ) TO gt_sort. APPEND VALUE #( field = 'CONNID' order = 'D' ) TO gt_sort. " 动态排序（复杂，需使用动态SQL或字段符号）

分组处理

" 1. AT NEW / AT END OF SORT gt_data BY carrid connid. LOOP AT gt_data INTO gs_data. AT NEW carrid. WRITE: / 'New Carrier:', gs_data-carrid. ENDAT. AT END OF connid. SUM. " 对数字字段求和 WRITE: / 'Sum for connection:', gs_data-connid. ENDAT. ENDLOOP. " 2. 使用REDUCE分组 DATA(lt_grouped) = REDUCE tt_sflight( INIT groups = VALUE tt_sflight( ) FOR GROUPS group OF wa IN gt_sflight GROUP BY ( carrier = wa-carrid ) ( VALUE #( BASE groups FOR <g> IN GROUP group ( <g> ) ) ) ).

五、内表函数与方法

5.1 DESCRIBE TABLE

DATA: lv_lines TYPE i, lv_kind TYPE abap_tablekind, lv_occurs TYPE i, lv_initial TYPE abap_bool. " 获取行数 DESCRIBE TABLE gt_table LINES lv_lines. " 获取表类型 DESCRIBE TABLE gt_table KIND lv_kind. " 获取初始大小 DESCRIBE TABLE gt_table OCCURS lv_occurs. " 是否为空 DESCRIBE TABLE gt_table LINES lv_lines. IF lv_lines = 0. " 空表 ENDIF. " 获取所有属性 DESCRIBE TABLE gt_table LINES lv_lines KIND lv_kind OCCURS lv_occurs.

5.2 LINES函数

" 获取行数（7.4+推荐） DATA(lv_count) = lines( gt_table ). " 在条件中使用 IF lines( gt_table ) > 0. " 表非空 ENDIF.

5.3 内表表达式函数

REDUCE

" 1. 求和 DATA(lv_total) = REDUCE i( INIT sum = 0 FOR wa IN gt_sflight NEXT sum = sum + wa-price ). " 2. 复杂聚合 TYPES: BEGIN OF ty_stats, carrier TYPE s_carr_id, count TYPE i, total TYPE s_price, avg TYPE s_price, END OF ty_stats. DATA(lt_stats) = REDUCE TABLE OF ty_stats( INIT result = VALUE tt_stats( ) FOR GROUPS group OF wa IN gt_sflight GROUP BY ( carrier = wa-carrid ) LET count = REDUCE i( INIT cnt = 0 FOR <g> IN GROUP group NEXT cnt = cnt + 1 ) total = REDUCE s_price( INIT t = 0 FOR <g> IN GROUP group NEXT t = t + <g>-price ) IN NEXT result = VALUE #( BASE result ( carrier = group-carrier count = count total = total avg = total / count ) ) ).

FILTER

" 过滤内表 DATA(lt_filtered) = FILTER #( gt_sflight WHERE carrid = 'LH' AND price > 1000 ). " 使用辅助键过滤 DATA(lt_by_price) = FILTER #( gt_sflight USING KEY price_key WHERE price > 500 ).

FOR表达式

" 生成新内表 DATA(lt_doubled) = VALUE tt_sflight( FOR wa IN gt_sflight ( CORRESPONDING #( BASE ( wa ) price = wa-price * 2 ) ) ). " 条件构造 DATA(lt_discounted) = VALUE tt_sflight( FOR wa IN gt_sflight LET discount = COND #( WHEN wa-price > 1000 THEN 0.9 WHEN wa-price > 500 THEN 0.95 ELSE 1 ) IN ( CORRESPONDING #( BASE ( wa ) price = wa-price * discount ) ) ).

六、性能优化深度指南

6.1 表类型选择策略

6.2 高效读取策略

" 1. 使用正确的查找方式 " 标准表：先SORT，再用BINARY SEARCH SORT gt_std BY carrid connid. READ TABLE gt_std INTO gs_line WITH KEY carrid = 'LH' connid = '0400' BINARY SEARCH. " 排序表：直接使用TABLE KEY READ TABLE gt_sorted INTO gs_line WITH TABLE KEY carrid = 'LH' connid = '0400'. " 哈希表：直接使用TABLE KEY READ TABLE gt_hashed INTO gs_line WITH TABLE KEY carrid = 'LH' connid = '0400'. " 2. 使用辅助键 " 声明时定义辅助键 DATA: gt_sorted TYPE SORTED TABLE OF sflight WITH NON-UNIQUE KEY primary_key COMPONENTS carrid connid WITH NON-UNIQUE SORTED KEY price_key COMPONENTS price. " 使用辅助键读取 READ TABLE gt_sorted INTO gs_line USING KEY price_key WITH KEY price = 1000. " 3. 避免全表扫描 " 错误：性能差 LOOP AT gt_table INTO gs_line. IF gs_line-carrid = 'LH'. " 处理 ENDIF. ENDLOOP. " 正确：使用WHERE或先过滤 LOOP AT gt_table INTO gs_line WHERE carrid = 'LH'. ENDLOOP. " 或使用FILTER DATA(lt_lh) = FILTER #( gt_table WHERE carrid = 'LH' ).

6.3 内存优化技巧

" 1. 设置合适的初始大小 DATA: gt_large TYPE TABLE OF sflight INITIAL SIZE 10000. " 2. 及时释放内存 FREE: gt_table. " 释放内存 CLEAR: gt_table. " 清空内容但保留结构 " 3. 使用PACKAGE SIZE分批处理 SELECT * FROM sflight INTO TABLE gt_buffer PACKAGE SIZE 1000. " 处理数据 CLEAR: gt_buffer. ENDSELECT. " 4. 避免深层结构 " 深层结构占用更多内存 TYPES: BEGIN OF ty_deep, header TYPE string, items TYPE TABLE OF ty_item, " 嵌套表 END OF ty_deep. " 5. 使用字段符号避免复制 FIELD-SYMBOLS: <fs_line> TYPE any. ASSIGN gt_table[ 1 ] TO <fs_line>. " 6. 删除不必要字段 " 读取时只取需要字段 SELECT carrid, connid, price FROM sflight INTO CORRESPONDING FIELDS OF TABLE gt_table WHERE carrid = 'LH'.

6.4 批量操作优化

" 1. 使用数组操作而非单行操作 " 差：逐行插入 LOOP AT gt_source INTO gs_source. INSERT gs_source INTO TABLE gt_target. ENDLOOP. " 好：批量插入 INSERT LINES OF gt_source INTO TABLE gt_target. " 2. 批量修改数据库 " 单行更新 LOOP AT gt_sflight INTO gs_sflight. UPDATE sflight FROM gs_sflight. ENDLOOP. " 批量更新 UPDATE sflight FROM TABLE gt_sflight. " 3. 使用FOR ALL ENTRIES优化 IF gt_keys IS NOT INITIAL. SELECT * FROM sflight INTO TABLE gt_data FOR ALL ENTRIES IN gt_keys WHERE carrid = gt_keys-carrid AND connid = gt_keys-connid. ENDIF.

七、高级主题

7.1 动态内表操作

" 1. 动态创建内表 DATA: lr_table TYPE REF TO data, lt_comp TYPE cl_abap_structdescr=>component_table. APPEND VALUE #( name = 'CARRID' type = CAST #( cl_abap_elemdescr=>get_c( p_length = 3 ) ) ) TO lt_comp. APPEND VALUE #( name = 'CONNID' type = CAST #( cl_abap_elemdescr=>get_n( p_length = 4 ) ) ) TO lt_comp. DATA(lo_struct) = cl_abap_structdescr=>create( lt_comp ). DATA(lo_table) = cl_abap_tabledescr=>create( lo_struct ). CREATE DATA lr_table TYPE HANDLE lo_table. ASSIGN lr_table->* TO FIELD-SYMBOL(<fs_dyn_table>). " 2. 动态读取 READ TABLE <fs_dyn_table> ASSIGNING FIELD-SYMBOL(<fs_dyn_line>) INDEX 1. " 3. 动态赋值 ASSIGN COMPONENT 'CARRID' OF STRUCTURE <fs_dyn_line> TO FIELD-SYMBOL(<fs_field>). <fs_field> = 'LH'.

7.2 内表与ALV集成

" 1. 准备内表数据 DATA: gt_sflight TYPE TABLE OF sflight, go_alv TYPE REF TO cl_salv_table. SELECT * FROM sflight INTO TABLE gt_sflight. " 2. 显示ALV cl_salv_table=>factory( IMPORTING r_salv_table = go_alv CHANGING t_table = gt_sflight ). go_alv->display( ). " 3. 设置ALV属性 DATA: lo_columns TYPE REF TO cl_salv_columns, lo_column TYPE REF TO cl_salv_column. lo_columns = go_alv->get_columns( ). lo_column = lo_columns->get_column( 'PRICE' ). lo_column->set_currency_column( 'CURRENCY' ).

7.3 内表序列化

" 1. 内表转XML DATA: lv_xml TYPE string. CALL TRANSFORMATION id SOURCE data = gt_sflight RESULT XML lv_xml. " 2. XML转内表 CALL TRANSFORMATION id SOURCE XML lv_xml RESULT data = gt_sflight. " 3. 内表转JSON（7.5+） DATA(lv_json) = /ui2/cl_json=>serialize( gt_sflight ). " 4. JSON转内表 /ui2/cl_json=>deserialize( EXPORTING json = lv_json CHANGING data = gt_sflight ).

7.4 内表调试技巧

" 1. 快速查看内表内容 " 在调试器中使用： " - 双击内表变量 " - 右键 → 显示为表 " - 使用内表查看器 " 2. 程序内输出内表 cl_demo_output=>display_data( gt_sflight ). " 3. 性能分析 " 使用事务码SAT或SE30 " 分析内表操作性能 " 4. 内存分析 " 使用事务码S_MEMORY_INSPECTOR

八、最佳实践与常见陷阱

8.1 最佳实践

" 1. 始终指定表类型和键 DATA: gt_table TYPE SORTED TABLE OF ty_structure WITH UNIQUE KEY field1 field2. " 2. 使用字段符号进行循环 LOOP AT gt_table ASSIGNING FIELD-SYMBOL(<fs_line>). <fs_line>-field = 'value'. ENDLOOP. " 3. 使用现代语法（7.4+） DATA(lt_filtered) = FILTER #( gt_table WHERE field = 'X' ). DATA(ls_line) = VALUE #( gt_table[ 1 ] OPTIONAL ). " 4. 批量操作数据库 UPDATE dbtable FROM TABLE gt_itab. " 5. 及时释放内存 FREE: gt_large_table.

8.2 常见陷阱

" 1. 忘记检查SY-SUBRC READ TABLE gt_table INTO gs_line WITH KEY field = 'X'. " 错误：未检查是否找到 gs_line-other_field = 'Y'. " 可能使用初始值 " 2. MODIFY WHERE修改多行 MODIFY gt_table FROM gs_line TRANSPORTING field WHERE key = 'X'. " 修改所有key='X'的行！ " 3. 内表复制错误 " 错误：浅拷贝 MOVE gt_table1 TO gt_table2. " 仅复制表头 " 正确：深拷贝 gt_table2 = gt_table1. " 4. FOR ALL ENTRIES为空 SELECT * FROM sflight INTO TABLE gt_result FOR ALL ENTRIES IN gt_keys WHERE carrid = gt_keys-carrid. " 如果gt_keys为空，会取出所有数据！ " 5. 循环中删除 LOOP AT gt_table INTO gs_line. IF gs_line-field = 'DELETE'. DELETE gt_table. " 错误！删除的是工作区，不是内表行 DELETE gt_table INDEX sy-tabix. " 正确 ENDIF. ENDLOOP.

8.3 性能检查清单

1. □选择了正确的表类型
2. □使用BINARY SEARCH前已排序
3. □避免嵌套循环，使用哈希表优化
4. □使用字段符号而非工作区
5. □批量操作而非逐行操作
6. □FOR ALL ENTRIES前检查内表非空
7. □及时释放大内表内存
8. □只读取需要的字段
9. □使用辅助键优化频繁查询
10. □避免在循环中排序

九、版本特性总结

十、实际应用示例

复杂业务逻辑处理

" 示例：销售订单处理 TYPES: BEGIN OF ty_order_summary, vbeln TYPE vbeln, kunnr TYPE kunnr, netwr TYPE netwr, currency TYPE waers, items TYPE i, delivery_date TYPE datum, priority TYPE char1, END OF ty_order_summary. " 处理逻辑 DATA(lt_summary) = REDUCE TABLE OF ty_order_summary( INIT result = VALUE tt_summary( ) FOR GROUPS order OF wa IN gt_vbap GROUP BY ( vbeln = wa-vbeln ) LET header = VALUE #( gt_vbek[ vbeln = order-vbeln ] OPTIONAL ) total_netwr = REDUCE netwr( INIT sum = 0 FOR item IN GROUP order NEXT sum = sum + item-netwr ) item_count = lines( FILTER #( gt_vbap WHERE vbeln = order-vbeln ) ) priority = COND #( WHEN total_netwr > 10000 THEN 'A' WHEN total_netwr > 5000 THEN 'B' ELSE 'C' ) IN NEXT result = VALUE #( BASE result ( vbeln = order-vbeln kunnr = header-kunnr netwr = total_netwr currency = header-waers items = item_count delivery_date = header-wadat_ist priority = priority ) ) ). " 按优先级排序 SORT lt_summary BY priority ASCENDING netwr DESCENDING.