CALL TRANSFORMATION: fix byte order mark (#901)

open-abap · Aug 7, 2024 · 2a4bdc9 · 2a4bdc9
1 parent 3ca1e63
commit 2a4bdc9
Show file tree

Hide file tree

Showing 3 changed files with 41 additions and 4 deletions.
diff --git a/src/conv/cl_abap_conv_in_ce.clas.abap b/src/conv/cl_abap_conv_in_ce.clas.abap
@@ -102,12 +102,15 @@ CLASS cl_abap_conv_in_ce IMPLEMENTATION.
 
     " Try TextDecoder first, if it runs in browser,
     WRITE '@KERNEL const decoder = TextDecoder || await import("util").TextDecoder;'.
-    WRITE '@KERNEL const td = new decoder(this.mv_js_encoding.get(), {fatal: this.mv_ignore_cerr.get() !== "X"});'.
+* https://stackoverflow.com/questions/62334608/textdecoder-prototype-ignorebom-not-working-as-expected
+    WRITE '@KERNEL const td = new decoder(this.mv_js_encoding.get(), {fatal: this.mv_ignore_cerr.get() !== "X", ignoreBOM: true});'.
     WRITE '@KERNEL try {'.
+    " WRITE '@KERNEL   console.dir(buf);'.
     WRITE '@KERNEL   data.set(td.decode(buf));'.
+    " WRITE '@KERNEL   console.dir(td.decode(buf).charCodeAt( 0 ));'.
     WRITE '@KERNEL } catch (e) {'.
-    WRITE '@KERNEL   console.dir(e);'.
-    WRITE '@KERNEL   console.dir(this.mv_js_encoding.get());'.
+*    WRITE '@KERNEL   console.dir(e);'.
+*    WRITE '@KERNEL   console.dir(this.mv_js_encoding.get());'.
     lv_error = abap_true.
     WRITE '@KERNEL }'.
 

diff --git a/src/kernel/call_transformation/kernel_call_transformation.clas.locals_imp.abap b/src/kernel/call_transformation/kernel_call_transformation.clas.locals_imp.abap
@@ -548,11 +548,22 @@ ENDCLASS.
 
 CLASS lcl_string_to_string IMPLEMENTATION.
   METHOD run.
+    DATA lv_str_bom TYPE string.
+    DATA lv_hex_bom TYPE xstring.
+
 * this is not right, but works for the unit test
     WRITE '@KERNEL result.set(INPUT.source.get());'.
 
+    lv_hex_bom = cl_abap_char_utilities=>byte_order_mark_little.
+    lv_str_bom = cl_abap_codepage=>convert_from(
+      source   = lv_hex_bom
+      codepage = 'UTF-16' ).
+
     IF options-xml_header = 'no'.
-      REPLACE FIRST OCCURRENCE OF REGEX '<?.*?>' IN result WITH ''.
+      REPLACE FIRST OCCURRENCE OF REGEX '<\?.*\?>' IN result WITH ''.
+      " WRITE '@KERNEL console.dir(lv_str_bom);'.
+      CONCATENATE lv_str_bom result INTO result.
+      " WRITE '@KERNEL console.dir(result);'.
     ENDIF.
   ENDMETHOD.
 ENDCLASS.
diff --git a/src/kernel/call_transformation/kernel_call_transformation.clas.testclasses.abap b/src/kernel/call_transformation/kernel_call_transformation.clas.testclasses.abap
@@ -128,6 +128,7 @@ CLASS ltcl_call_transformation DEFINITION FOR TESTING RISK LEVEL HARMLESS DURATI
     METHODS suppress3 FOR TESTING RAISING cx_static_check.
     METHODS xml_to_xml FOR TESTING RAISING cx_static_check.
     METHODS xml_to_xml_rm_header FOR TESTING RAISING cx_static_check.
+    METHODS xml_to_xml_rm_header_bom FOR TESTING RAISING cx_static_check.
 ENDCLASS.
 
 CLASS ltcl_call_transformation IMPLEMENTATION.
@@ -1119,4 +1120,26 @@ CLASS ltcl_call_transformation IMPLEMENTATION.
 
   ENDMETHOD.
 
+  METHOD xml_to_xml_rm_header_bom.
+
+    DATA lv_xml     TYPE string.
+    DATA lv_str_bom TYPE string.
+    DATA lv_hex_bom TYPE xstring.
+
+    lv_xml = |<foo>2</foo>|.
+    CALL TRANSFORMATION id SOURCE XML lv_xml RESULT XML lv_xml OPTIONS xml_header = 'no'.
+
+    lv_hex_bom = cl_abap_char_utilities=>byte_order_mark_little.
+    lv_str_bom = cl_abap_codepage=>convert_from(
+      source   = lv_hex_bom
+      codepage = 'UTF-16' ).
+
+    lv_xml = lv_xml(1).
+
+    cl_abap_unit_assert=>assert_equals(
+      act = lv_xml
+      exp = lv_str_bom ).
+
+  ENDMETHOD.
+
 ENDCLASS.