如何基于ID及Origin多条件对ABAP内部表进行分类标记？

输入内部表 #

期望结果 #

问题说明 #

现有上述ABAP内部表，需按ID分组生成统一的Classification字段：

• 前缀根据ID下不同Origin的数量确定：3种对应Triple，2种对应Double，其他情况对应Single
• 后缀为该ID下所有不同Origin的下划线拼接字符串
  尝试循环筛选ID并检查Origin值未实现目标，寻求更优方案（包括SELECT语句处理方式）

用户尝试代码 #

DATA: lv_current_id   TYPE string VALUE ' ',
      lv_origin_count TYPE i VALUE 0,
      lv_origin_text  TYPE string VALUE ' '.

LOOP AT lt_internal_table INTO DATA(ls_data).

  CLEAR: lv_origin_count, lv_origin_text.

  " Check, if origin is processed
  IF NOT ( ls_data-origin IN lt_origin ).
    " Save origin in temporary list
    APPEND ls_data-origin TO lt_origin.

    " Count occurrence of origin per id
    LOOP AT lt_data INTO DATA(ls_data2) WHERE id = ls_data-id AND origin = ls_data-origin.
      ADD 1 TO lv_origin_count.
    ENDLOOP.

    " Create cext for classification columnText für die Anzahl_Herkunft erstellen
    CASE lv_origin_count.
      WHEN 1.
        lv_origin_text = 'Single' && ls_data-origin.
      WHEN 2.
        lv_origin_text = 'Double' && ls_data-origin && ls_data-origin.
      WHEN 3.
        lv_origin_text = 'Triple' && ls_data-origin && ls_data-origin && ls_data-origin.
    ENDCASE.

    " Save result in internal result table 
    APPEND VALUE #( id = ls_data-id text = ls_data-text origin = ls_data-origin classification = lv_origin_text ) TO lt_result.
  ENDIF.
ENDLOOP.

解决方案 #

方案一：ABAP内部表循环处理（适用于所有版本） #

通过两次循环完成分组统计与结果生成，逻辑清晰且兼容低版本ABAP：

* 定义类型
TYPES: BEGIN OF ty_source,
         id        TYPE i,
         text      TYPE string,
         origin    TYPE string,
       END OF ty_source.

TYPES: BEGIN OF ty_id_meta,
         id        TYPE i,
         origins   TYPE STANDARD TABLE OF string WITH UNIQUE KEY table_line,
         class     TYPE string,
       END OF ty_id_meta.

* 声明变量
DATA: lt_source     TYPE STANDARD TABLE OF ty_source,
      lt_id_metas   TYPE STANDARD TABLE OF ty_id_meta WITH UNIQUE KEY id,
      lt_result     TYPE STANDARD TABLE OF ty_source WITH EMPTY KEY.

* 假设lt_source已填充输入数据

* 第一步：按ID收集所有唯一Origin
LOOP AT lt_source INTO DATA(ls_source).
  READ TABLE lt_id_metas INTO DATA(ls_id_meta) WITH KEY id = ls_source-id.
  IF sy-subrc <> 0.
    ls_id_meta-id = ls_source-id.
    INSERT ls_source-origin INTO ls_id_meta-origins INDEX 1.
    INSERT ls_id_meta INTO lt_id_metas.
  ELSE.
    " 仅添加未存在的Origin
    READ TABLE ls_id_meta-origins TRANSPORTING NO FIELDS WITH KEY table_line = ls_source-origin.
    IF sy-subrc <> 0.
      INSERT ls_source-origin INTO ls_id_meta-origins INDEX 1.
      MODIFY lt_id_metas FROM ls_id_meta INDEX sy-tabix.
    ENDIF.
  ENDIF.
ENDLOOP.

* 第二步：生成每个ID对应的Classification
LOOP AT lt_id_metas INTO ls_id_meta.
  SORT ls_id_meta-origins. " 确保Origin拼接顺序一致
  DATA(lv_origin_count) = lines(ls_id_meta-origins).
  " 拼接Origin字符串
  DATA(lv_origin_str) = REDUCE string( 
    INIT res = '' 
    FOR <origin> IN ls_id_meta-origins 
    NEXT res = res && COND( WHEN res IS NOT INITIAL THEN '_' ) && <origin> 
  ).
  " 生成分类前缀
  CASE lv_origin_count.
    WHEN 2.
      ls_id_meta-class = `Double_` && lv_origin_str.
    WHEN 3.
      ls_id_meta-class = `Triple_` && lv_origin_str.
    WHEN OTHERS.
      ls_id_meta-class = `Single_` && lv_origin_str.
  ENDCASE.
  MODIFY lt_id_metas FROM ls_id_meta INDEX sy-tabix.
ENDLOOP.

* 第三步：关联原表生成最终结果
LOOP AT lt_source INTO ls_source.
  READ TABLE lt_id_metas INTO ls_id_meta WITH KEY id = ls_source-id.
  APPEND VALUE #(
    id = ls_source-id
    origin = ls_source-origin
    classification = ls_id_meta-class
  ) TO lt_result.
ENDLOOP.

方案二：SELECT语句处理（适用于ABAP 7.50+，数据来自数据库表） #

利用ABAP 7.50支持的STRING_AGG函数直接在数据库层面完成分组与字符串聚合，性能更优：

SELECT id,
       origin,
       CASE COUNT(DISTINCT origin)
         WHEN 2 THEN 'Double_' || STRING_AGG( DISTINCT origin, '_' ORDER BY origin )
         WHEN 3 THEN 'Triple_' || STRING_AGG( DISTINCT origin, '_' ORDER BY origin )
         ELSE 'Single_' || STRING_AGG( DISTINCT origin, '_' ORDER BY origin )
       END AS classification
FROM your_database_table " 替换为实际数据库表名
GROUP BY id, origin
ORDER BY id, origin.

内容的提问来源于stack exchange，提问作者christhebliss