青海网站开发今日新闻简报

一、前言

目前项目中很多地方会用到将ABAP的格式转换为JSON或者XML格式，下面介绍几种方法：

對transformation的方式生成XML文件，方法有多種：

• call transformation 
• cl_proxy_xml_transform=>abap_to_xml_xstring的方式：
• /ui2/cl_json
• CL_TREX_JSON_SERIALIZER与CL_TREX_JSON_DESERIALIZER

二、call transformation

  DATA: lo_writer TYPE REF TO cl_sxml_string_writer.lo_writer = cl_sxml_string_writer=>create(type                     = if_sxml=>co_xt_xml10encoding                 = `UTF-8` ).CALL TRANSFORMATION idSOURCE spfli = lt_itabsflight = lt_itab2RESULT XML lo_writer.lv_xml = lo_writer->get_output( ).CALL FUNCTION 'SCMS_XSTRING_TO_BINARY'EXPORTINGbuffer        = lv_xmlIMPORTINGoutput_length = lv_sizeTABLESbinary_tab    = lt_xml.

若是需要json格式，則可以將type的參數改為IF_SXML=>CO_XT_BINARY，若目標內表中存在非扁平結構，需要在SE11中提前定義好結構，若需要像PI結構一樣可以控制initial字段，內表在最終的XML或者文件中可見，則需要在SE80中定義結構Enterprise Services中定義。

三、cl_proxy_xml_transform=>abap_to_xml_xstring

  cl_proxy_xml_transform=>abap_to_xml_xstring(EXPORTINGabap_data    = is_itab3ddic_type    = 'ZLIAN_TEST_XML' " Name of Dictionary Type
*      ext_xml      = ABAP_FALSE
*      xml_header   = 'no'
*      root_element =
*      svar_name    =     " Proxy Generation: Name of an Object in R/3RECEIVINGxml          = lv_xml ).CATCH cx_proxy_fault.    " Proxy FaultCATCH cx_transformation_error.    " General Error When Performing CALL TRANSFORMATION

注：若是多層結構，兩種方式需要建立的結構不一樣，cl_proxy_xml_transform=>abap_to_xml_xstring不會產生最內層的tablel_line，最外層的name也不會產生；也不會帶出空字段，call transformation則都相反。

四、/ui2/cl_json

JSON转 ABAP结构

TYPES: BEGIN OF t_record,key1 TYPE string,key2 TYPE string,END OF t_record.DATA: json TYPE string.
DATA: record TYPE t_record.json = '{"key1": "VALUE1",  "key2": "VALUE2"}'./ui2/cl_json=>deserialize( EXPORTING json = json CHANGING data = record ).
IF sy-subrc = 0. ENDIF.

ABAP结构转JSON

json = /ui2/cl_json=>serialize( data = record compress = abap_true pretty_name = /ui2/cl_json=>pretty_mode-camel_case ).

当然SAP还是提供了其它类来进行JSON格式的转换。

但是这种JSON的方法出来的名字会直接保存为小写，意味着KEY1/KEY2在JSON中永远都是key1/key2

{"key1":"ValuE1","key2":"VALUE2"}

怎样才可以做到{"KEy1":"ValuE1","key2":"VALUE2"}

这时候需要做到驼峰美化

DATA: BEGIN OF record,_K_EY1 TYPE string VALUE 'ValuE1',key2 TYPE string VALUE 'VALUE2',END OF record.
DATA: json TYPE string.json = /ui2/cl_json=>serialize( data = record compress = abap_true pretty_name = /ui2/cl_json=>pretty_mode-camel_case ).

五、CL_TREX_JSON_SERIALIZER与CL_TREX_JSON_DESERIALIZER

• CL_TREX_JSON_SERIALIZER ：将 ABAP 内表转换为 json 格式
• CL_TREX_JSON_DESERIALIZER：json 转换为 ABAP 内表

这两个类有一点点问题，序列化生成的格式 key 没有引号，可能导致外部解析的失败。比如

[{id: "001",name: "Stone"},{id: "002",name: "Brown"}
]

 而我们需要的是下面的格式：

[{"id": "001","name": "Stone"},{"id": "002","name": "Brown"}
]

所以需要对两个类进行改造，以符合自己的需求。首先我们使用事务码 SE24 将 CL_TREX_JSON_SERIALIZER 类拷贝一个新类，另存为 ZCL_TREX_JSON_SERIALIZER，然后将 RECURSE() 方法做两点改变：

填加语句：

注释掉 CONCATENATE <abapcomp>-name c_colon INTO l_value 语句，改写为下面的语句。就是在 <abapcomp>-name 两边加上双引号 ：

 

完整的 RECURSE() 方法代码如下：

METHOD recurse.DATA:l_type  TYPE c ,l_comps TYPE i ,l_lines TYPE i ,l_index TYPE i ,l_value TYPE string .FIELD-SYMBOLS:<itab> TYPE ANY TABLE ,<comp> TYPE ANY .DESCRIBE FIELD data TYPE l_type COMPONENTS l_comps .IF l_type = cl_abap_typedescr=>typekind_table .
*   itab -> arrayAPPEND '[' TO me->fragments .ASSIGN data TO <itab> .l_lines = LINES( <itab> ) .LOOP AT <itab> ASSIGNING <comp> .ADD 1 TO l_index .recurse( <comp> ) .IF l_index < l_lines .APPEND c_comma TO me->fragments .ENDIF .ENDLOOP .APPEND ']' TO fragments .ELSE .IF l_comps IS INITIAL .
*     field -> scalar
*     todo: formatl_value = data .REPLACE ALL OCCURRENCES OF '\' IN l_value WITH '\\' .REPLACE ALL OCCURRENCES OF '''' IN l_value WITH '\''' .REPLACE ALL OCCURRENCES OF '"' IN l_value WITH '\"' .REPLACE ALL OCCURRENCES OF '&' IN l_value WITH '\&' .REPLACE ALL OCCURRENCES OF cl_abap_char_utilities=>cr_lf IN l_value WITH '\r\n' .REPLACE ALL OCCURRENCES OF cl_abap_char_utilities=>newline IN l_value WITH '\n' .REPLACE ALL OCCURRENCES OF cl_abap_char_utilities=>horizontal_tab IN l_value WITH '\t' .REPLACE ALL OCCURRENCES OF cl_abap_char_utilities=>backspace IN l_value WITH '\b' .REPLACE ALL OCCURRENCES OF cl_abap_char_utilities=>form_feed IN l_value WITH '\f' .CONDENSE l_value.CONCATENATE '"' l_value '"' INTO l_value .APPEND l_value TO me->fragments .ELSE .
*     structure -> objectDATA l_typedescr TYPE REF TO cl_abap_structdescr .FIELD-SYMBOLS <abapcomp> TYPE abap_compdescr .APPEND '{' TO me->fragments .l_typedescr ?= cl_abap_typedescr=>describe_by_data( data ) .LOOP AT l_typedescr->components ASSIGNING <abapcomp> .l_index = sy-tabix .
*        CONCATENATE <abapcomp>-name c_colon INTO l_value .CONCATENATE '"' <abapcomp>-name '"' c_colon INTO l_value.TRANSLATE l_value TO LOWER CASE .APPEND l_value TO me->fragments .ASSIGN COMPONENT <abapcomp>-name OF STRUCTURE data TO <comp> .recurse( <comp> ) .IF l_index < l_comps .APPEND c_comma TO me->fragments .ENDIF .ENDLOOP .APPEND '}' TO me->fragments .ENDIF .ENDIF .ENDMETHOD.

 相对应地， 将 CL_TREX_JSON_DESERIALIZER 拷贝成为 ZCL_TREX_JSON_DESERIALIZER，并且对 deserialize_object 方法做两点变更 (网上有一个做好的图片，我就直接贴图了)：

完成后，deserialize_object 方法的完整代码如下：

METHOD deserialize_object.DATA:l_node_type TYPE REF TO cl_abap_typedescr ,l_ref TYPE REF TO object .ADD 1 TO offset . "skip {l_node_type = cl_abap_typedescr=>describe_by_data( node ) .* prepare for dynamic accessCASE l_node_type->kind .WHEN cl_abap_typedescr=>kind_ref .l_ref = node .WHEN cl_abap_typedescr=>kind_struct .WHEN OTHERS .RAISE EXCEPTION TYPE cx_trex_serialization .ENDCASE .DATA:l_done TYPE abap_bool ,l_len TYPE i ,l_name TYPE string .* handle each componentWHILE l_done = abap_false ."find next keyFIND REGEX '"(\w+)\s*":' IN SECTION OFFSET offset OF jsonMATCH OFFSET offset MATCH LENGTH l_lenSUBMATCHES l_name .IF sy-subrc <> 0 .RAISE EXCEPTION TYPE cx_trex_serialization .ENDIF .ADD l_len TO offset .FIELD-SYMBOLS <comp> TYPE ANY .*   dynamic binding to componentTRANSLATE l_name TO UPPER CASE .CASE l_node_type->kind .WHEN cl_abap_typedescr=>kind_ref .ASSIGN l_ref->(l_name) TO <comp> .WHEN cl_abap_typedescr=>kind_struct .ASSIGN COMPONENT l_name OF STRUCTURE node TO <comp> .IF sy-subrc <> 0.CONTINUE.ENDIF.WHEN OTHERS .RAISE EXCEPTION TYPE cx_trex_serialization .ENDCASE .DATA:l_comp_type TYPE REF TO cl_abap_typedescr ,l_ref_type TYPE REF TO cl_abap_refdescr .*   check component typel_comp_type = cl_abap_typedescr=>describe_by_data( <comp> ) .CASE l_comp_type->kind .
*     create instance if it's an orefWHEN cl_abap_typedescr=>kind_ref .l_ref_type ?= l_comp_type .l_comp_type = l_ref_type->get_referenced_type( ) .CREATE OBJECT <comp> TYPE (l_comp_type->absolute_name) .ENDCASE .*   deserialize current componentdeserialize_node(EXPORTINGjson = jsonCHANGINGoffset = offsetnode = <comp> ) .FIND REGEX ',|\}' IN SECTION OFFSET offset OF json MATCH OFFSET offset .IF sy-subrc <> 0 .RAISE EXCEPTION TYPE cx_trex_serialization .ENDIF .IF json+offset(1) = '}' .l_done = abap_true .ENDIF .ADD 1 TO offset .ENDWHILE .ENDMETHOD.

 示例

调用 ZCL_TREX_JSON_SERIALIZER 实现序列化以及调用 ZCL_TREX_JSON_DESERIALIZER 实现反序列化的代码如下：

REPORT  zabap_json .DATA: serializer TYPE REF TO zcl_trex_json_serializer,lv_json TYPE string.DATA: BEGIN OF ls_json,id TYPE string,name TYPE string,END OF ls_json.DATA: lt_json LIKE STANDARD TABLE OF ls_json.CLEAR ls_json.
ls_json-id = '001'.
ls_json-name = 'Stone'.APPEND ls_json TO lt_json.CLEAR ls_json.
ls_json-id = '002'.
ls_json-name = 'Brown'.APPEND ls_json TO lt_json.CREATE OBJECT serializerEXPORTING DATA = lt_json[].
CALL METHOD serializer->serialize( ).
lv_json = serializer->get_data( ).WRITE / lv_json.* DeserialzeWRITE /.
WRITE: / 'Deseriaze json string to internal table: '.DATA: deserializer TYPE REF TO zcl_trex_json_deserializer.
CLEAR lt_json.CREATE OBJECT deserializer.
CALL METHOD deserializer->deserialize(EXPORTING json = lv_jsonIMPORTING abap = lt_json[] ).LOOP AT lt_json INTO ls_json.WRITE : / ls_json-id, ls_json-name.
ENDLOOP.