Unicode printing problem (specifically xdot)

[birgerro]

Some unicode characters, like ẋ (xdot), ẍ (xdotdot) and ẏ (ydot), show up correctly on screen, but are replaced with a squares when printing the included python file. Strangly enough, this does not happen all the time, but only on some lines.

xdot_test.pdf

xdot_test.py

Actually, it seems ẋ prints correctly when there is a two-byte character z̈ (zdotdot, \u007a\u0308) on the same line, but is replaced by a square when z̈ is absent.

xdot_test2.pdf
xdot_test2.py

[hpwamr]

HI @RaiKoHoff ,
I confirm the bad printing within Notepad3. 🤔
Printing within Notepad++ is OK.

[RaiKoHoff]

Claude Code Analysis:

This isn't character scrambling, it's missing-glyph boxes (so-called "tofu" — the .notdef rectangle GDI emits when a font has no glyph for a code point):

  ┌───────┬──────────────────────────────────────────────┬──────────────┐
  │ Char  │                  Code point                  │    In PDF    │
  ├───────┼──────────────────────────────────────────────┼──────────────┤
  │ ẋ ẍ ẏ │ U+1E8B / 8D / 8F (Latin Extended Additional) │ ▯ tofu      │
  ├───────┼──────────────────────────────────────────────┼──────────────┤
  │ ₖ     │ U+2096 (Subscripts)                          │ ▯ tofu      │
  ├───────┼──────────────────────────────────────────────┼──────────────┤
  │ ÿ     │ U+00FF (Latin-1)                             │ renders fine │
  ├───────┼──────────────────────────────────────────────┼──────────────┤
  │ ż     │ U+017C (Latin Extended-A)                    │ renders fine │
  └───────┴──────────────────────────────────────────────┴──────────────┘

The screen renders all of them correctly because Notepad3's editor uses DirectWrite (SC_TECHNOLOGY_DIRECTWRITERETAIN),
which automatically falls back to other system fonts for code points the primary font doesn't cover.

The print path cannot use DirectWrite — printer DCs only support GDI — so Scintilla forces Technology::Default for the
print surface (scintilla\src\EditView.cxx:2685). SurfaceGDI::DrawTextCommonUTF8 calls ExtTextOutW directly
(scintilla\win32\SurfaceGDI.cxx:755) with no Uniscribe shaping and no font-linking — so any code point not present in the
configured editor font is replaced by .notdef on the printer.

The pattern in your PDF is consistent with the editor font being something like Consolas or Lucida Console: those cover
Latin-1 (ÿ) and Latin Extended-A (ż), but don't ship glyphs for the subscripts block (ₖ) or all of Latin Extended
Additional (ẋ/ẍ/ẏ). DirectWrite hides this on screen by silently substituting from another font.

Why this isn't a quick code fix in Notepad3:
The fix has to live in Scintilla's SurfaceGDI (Uniscribe-based shaping with font linking, or rendering via Direct2D into
a bitmap that's then BitBlt'd to the printer DC). Either is a substantial rework. The Notepad3 Print.cpp pipeline is
correct.

Proper fix: port a SurfaceGDI Uniscribe shaping path (or D2D-print path) — significantly bigger, upstream-Scintilla
territory.

Compare with Notepad++ (another Scintilla based editor)

[birgerro]

Could you also ask Claude Code why ẋ, ẍ and ẏ is printed correctly when there is a two-byte character z̈ on the same line? That seems like a quick fix (or probably a hack): Include a (tine, background colored) z̈ following each ẋ (or something similar).

[RaiKoHoff]

Rather than implementing a tactical patch for this individual case, we propose addressing the more general problem at its source. This will ensure greater stability and prevent technical debt moving forward.

As this issue originates within the upstream source, we recommend utilizing an alternative font that offers better compatibility with the printing hardware as a primary workaround.

[birgerro]

That sounds like a good solution. Do you have any idea about alternative fonts with better compatibility for the subscript block and Latin Extended Additional? I have been using the defaults for Notepad3:

BASE (Default Style): Cascadia Code / Regular / Size 11
and the same for Python Script: Default Style

Edit: Just to answer my own question: The font Calibri did a good job for my set of unicode characters.

[hpwamr]

The font Calibri did a good job for my set of unicode characters.

Please note that the font "Calibri" is a "Proportional" font.
I advise against using it with Notepad3 and the coding's lexers (because the columns are not vertically aligned). 🧐

In my list of Monospace fonts, only DejaVu Sans Mono works correctly with your text (including the small subscript "k").

   

PS: To access the list of "Monospace" fonts installed on your system, hold down the Shift key when you open the lists of "Global Default Font..." or "Current Scheme's Default Fonts..."