Default template theme forces East-Asian text to fall back to MS Mincho (ＭＳ 明朝)

[sebastEXlabe]

Default template's theme forces East-Asian text to fall back to MS Mincho (ＭＳ 明朝)

Summary

Document() (a fresh document created from the bundled default.docx template) configures the
theme and document settings so that East-Asian text resolves to the Japanese font ＭＳ 明朝 (MS Mincho),
even when the user never asked for a Japanese font and even after setting an explicit East-Asian
font with run.font.name. On systems that don't have MS Mincho installed (a standard Chinese/English
Windows), the CJK glyphs are then substituted / rendered in the wrong typeface.

This is distinct from the long-standing API gaps in #346 / #154 / #1396 (which are about
Font.name not writing the w:eastAsia attribute). Those reports are about the user-side setter.
This issue is about the default template itself shipping a mis-configured theme that make the
fallback land on a Japanese font.

Environment

• python-docx: 1.2.0 (verified against master, byte-identical template)
• Word / WPS on a Chinese-locale Windows without the MS Mincho / MS Gothic fonts installed

Reproduction

from docx import Document

doc = Document()          # fresh document from bundled default.docx
p = doc.add_paragraph()
run = p.add_run("中文字体回退测试 Chinese 123")
# Default behavior - we never set any font.
doc.save("repro.docx")

The run python-docx emits

<w:r><w:t>中文字体回退测试 Chinese 123</w:t></w:r>

There is no <w:rPr>/<w:rFonts> on the run at all. So the CJK glyphs must inherit from the theme.

Root cause chain (all confirmed from the shipped template)

The inherited East-Asian font resolves through this chain, and every link is wrong:

1. word/settings.xml — themeFontLang declares the document's East-Asian language as Japanese:

   <w:themeFontLang w:val="en-US" w:eastAsia="ja-JP"/>

2. word/theme/theme1.xml — the minorFont (used by Normal text via docDefaults) has an empty
   East-Asian and complex-script typeface, and its Jpan script font is a Japanese face:

   <a:minorFont>
     <a:latin typeface="Cambria"/>
     <a:ea typeface=""/>
     <a:cs typeface=""/>
     <a:font script="Jpan" typeface="ＭＳ 明朝"/>
     ...
   </a:minorFont>

   (majorFont has the same empty ea/cs with Jpan = ＭＳ ゴシック.)

3. word/styles.xml — docDefaults references the theme East-Asian font instead of a concrete font:

   <w:rFonts w:asciiTheme="minorHAnsi" w:eastAsiaTheme="minorEastAsia" w:hAnsiTheme="minorHAnsi" w:cstheme="minorBidi"/>

4. word/fontTable.xml — the theme declares the Japanese faces ＭＳ 明朝 and ＭＳ ゴシック as
   real fonts (charset 0x80), so they are treated as available.

Resolution: run has no font → docDefaults gives eastAsiaTheme="minorEastAsia" → theme minorFont
East-Asian typeface is empty → because themeFontLang eastAsia is ja-JP, the document's East-Asian
language is Japanese → the Jpan script font (ＭＳ 明朝) is selected.

Even setting run.font.name = '宋体' does NOT fix the CJK glyphs, because python-docx only writes
w:ascii/w:hAnsi (see #346 / #154) and the East-Asian characters still fall through to the theme's
Japanese font.

Expected behavior

• A freshly created Document() should not, by default, resolve CJK text to a Japanese typeface.
• The default template should either declare a neutral/Chinese East-Asian language
  (w:eastAsia="zh-CN"), or specify a concrete, localizable East-Asian font in the theme's
  a:ea/a:cs slots, so CJK glyphs don't silently fall back to MS Mincho.

Possible fix (in the bundled template)

• settings.xml: w:themeFontLang ... w:eastAsia="zh-CN" (or drop the Japanese default).
• theme1.xml: give a:ea/a:cs a non-empty, meaningful typeface (e.g. 宋体/SimSun), or point
  Jpan at a font that shippers commonly have; or set these to a neutral theme-resolvable value.

Related, non-duplicate issues

• feature: Font.name_far_east #154 — feature request for Font.name_far_east (API does not expose East-Asian font setter)
• Cannot set Chinese character font typeface #346 — run.font.name does not apply to Chinese characters (same API limitation)
• Chinese fonts Only the non-Chinese parts are valid #1396 — confirms run.font.name writes only ascii/hAnsi, never eastAsia

These describe the missing eastAsia API; none of them reports that the default template itself
ships a Japanese-language theme that makes the fallback land on MS Mincho.

[lsaint]

This is one of the gaps I ran into as well. I maintain python-docx-oss, which keeps the same docx API but exposes the font slots separately, including run.font.eastAsia, so CJK fonts can be set without dropping down to raw XML.

It doesn’t address the bundled template/theme issue you identified here, but it may be useful as an immediate workaround for the missing East Asian font setter:
https://github.com/lsaint/python-docx-oss