We convert XLSX to PDF with the following call, and pass no other parameters.
PUT /v3.0/cells/convert?format=PDF
Aspose.Cells Cloud 26.7.0
The PDFs we examined were produced on 2026-07-03 with v26.6, and we reproduced the
same result on 2026-07-30 with v26.7 (times are JST, UTC+9).
Problem
Myanmar and Devanagari text is not shaped. Characters that should combine into a
single glyph are drawn separately, and invisible marks such as U+1039 and U+094D
are drawn as visible glyphs.
Example: လက္ခဏာ
Expected: ခ stacked under က, and U+1039 not rendered at all
Actual: three separate glyphs, with U+1039 visible as a “+” shaped mark
We decoded the drawn glyphs in the output PDF through the ToUnicode CMaps and
found no ligature glyphs at all.
Myanmar: 14 glyphs map to U+1039 alone, 0 ligature glyphs
Nepali: 221 glyphs map to U+094D alone, 0 ligature glyphs
Reordering is missing as well. The Devanagari i-matra (U+093F, 94 occurrences) is
drawn after its consonant instead of before it.
The fonts are not the problem. Myanmar Text and Nirmala UI are correctly embedded
in the output PDF, and the same strings render correctly in Chrome.
We also tried
- Converting XLSX to DOCX with Aspose.Cells, then DOCX to PDF with Aspose.Words.
The result was the same. - Looking for a typography option in the API specification. We found none that
applies to PDF output. IsFontSubstitutionCharGranularity, DefaultFont and
DefaultEditLanguage exist only in ImageOrPrintOptions, which the PDF conversion
does not accept, and there is no PdfSaveOptions definition.
How to reproduce
Put Myanmar or Devanagari text into a cell or a text box in an XLSX and convert it
with the call above. Any text containing a virama is enough.
Myanmar လက္ခဏာ = U+101C U+1000 U+1039 U+1001 U+100F U+102C
Devanagari शिक्षा = U+0936 U+093F U+0915 U+094D U+0937 U+093E
In the resulting PDF, U+1039 and U+094D appear as separate visible glyphs and no
ligature is formed. We can send the source file and the produced PDF on request.
Questions
- Does Aspose.Cells Cloud support complex script shaping for Myanmar and
Devanagari when rendering to PDF? - If it does, what are we missing?
shaping-comparison.png (186.6 KB)