Writing Large Files (Streaming Writer)

zcl_excel_writer_huge_file is a streaming XSLT-based writer designed for workbooks that are too large to build entirely in memory with zcl_excel_writer_2007. It generates the XML directly to a string or file stream, bypassing the in-memory cell table, and is therefore suitable for hundreds of thousands of rows.

When to Use

Scenario	Recommended writer
< ~50 000 rows, full formatting, charts, images	zcl_excel_writer_2007
50 000 – 1 000 000+ rows, minimal formatting	zcl_excel_writer_huge_file
CSV export	zcl_excel_writer_csv

Trade-offs: zcl_excel_writer_huge_file supports cell values, basic number formats, and simple styles. It does not support charts, images, drawings, named ranges, or complex merged-cell layouts. Use it only for high-volume tabular data exports.

Basic Usage

DATA: lo_excel  TYPE REF TO zcl_excel,
      lo_ws     TYPE REF TO zcl_excel_worksheet,
      lo_writer TYPE REF TO zcl_excel_writer_huge_file.

CREATE OBJECT lo_excel.
lo_ws = lo_excel->get_active_worksheet( ).
lo_ws->set_title( 'Data Export' ).

" Write cells as normal — the streaming writer collects cell data per-row
DATA(lv_row) = 1.
lo_ws->set_cell( ip_column = 'A' ip_row = lv_row ip_value = 'OrderNo' ).
lo_ws->set_cell( ip_column = 'B' ip_row = lv_row ip_value = 'Customer' ).
lo_ws->set_cell( ip_column = 'C' ip_row = lv_row ip_value = 'Amount' ).

LOOP AT lt_orders INTO DATA(ls_order).
  ADD 1 TO lv_row.
  lo_ws->set_cell( ip_column = 'A' ip_row = lv_row ip_value = ls_order-order_no ).
  lo_ws->set_cell( ip_column = 'B' ip_row = lv_row ip_value = ls_order-customer ).
  lo_ws->set_cell( ip_column = 'C' ip_row = lv_row ip_value = ls_order-amount ).
ENDLOOP.

" Use the huge-file writer instead of zcl_excel_writer_2007
CREATE OBJECT lo_writer.
DATA(lv_file) = lo_writer->write_file( lo_excel ).

The return type of write_file is xstring, identical to zcl_excel_writer_2007.

Using bind_table with the Streaming Writer

The most efficient pattern for large datasets is bind_table followed by the huge-file writer. bind_table populates the in-memory cell table row-by-row from an internal table:

" Populate worksheet from an internal table
lo_ws->bind_table( ip_table = lt_large_data ).

" Stream it out
CREATE OBJECT lo_writer.
DATA(lv_xstring) = lo_writer->write_file( lo_excel ).

Memory-Saving Tips

1. Process data in chunks. If the source data is read from the database, use PACKAGE SIZE to read and write in batches, clearing the previous batch:

   SELECT * FROM zsales_data INTO TABLE @DATA(lt_chunk) PACKAGE SIZE 10000.
     lo_ws->bind_table( ip_table = lt_chunk ip_start_row = lv_next_row ).
     lv_next_row = lv_next_row + lines( lt_chunk ).
   ENDSELECT.

2. Avoid complex styles in loops. Every distinct style creates a new entry in the workbook's style table. Create styles once before the loop and reuse the reference.

3. Use bind_table over set_cell loops where possible — bind_table is significantly faster for bulk population.

Limitations

• No charts, images, or drawings
• No named ranges or defined names
• No complex merged cells (header merges before the data are OK)
• No OLE automation features (use zcl_excel_ole for that — classic ABAP only)
• Thread safety: the writer is stateless; one instance per write_file call is safe

Comparing the Three Writers

Feature	writer_2007	writer_huge_file	writer_csv
Full cell styles	✅	⚠️ Basic	❌
Charts / images	✅	❌	❌
Named ranges	✅	❌	❌
Page setup / print	✅	⚠️ Limited	❌
AutoFilter	✅	❌	via is_row_hidden
Structured tables	✅	❌	❌
Row limit (practical)	~50 000	1 000 000+	Unlimited
Output format	.xlsx	.xlsx	.csv
Cloud-compatible	✅	✅	✅

Next Steps

• Data Conversion — bind_table reference
• CSV Export — for text-only exports
• Cloud Compatibility — verify your system supports streaming