Use multiple passes to separate fonts and "kerning". (This can be done through PUA."
to\uEXXX
to<span>"</span>
.)Use only left-left, left-right (and right-right?) to deal with consec puncts.Preserve the tables drawing symbols in SimSun/pre.Add a "preserved Tags" class.code
,tt
are generally safe but the format should be preserved.- Change italic CJK to other fonts.
Fix the "SAFED by USER" rule.Fix the "nomore spaces rule" on some pages,e.g. sciencenet.cn.- Fix the "dangling punctuations".
- Add
​
before and after slash? - Add a "half-safe" class for
div
? - Add variable length space to the end of each paragraph? (seems no harm to add a
br
beforep
?) - Also encapsulate Latin QM if context is in SimSun?
- Use BFS to traverse the DOM tree?
Fix the hidden elements problem.- Optimize speed according to HTMLElement.lang
- Defer label "lang=en,..." until double click.
- Use google fonts service to fetch online fonts.
- Fix through-tag auto-space when the parent node contains no CJK or puncts.