You can display 格
using the numeric entity reference 格, etc. The encoding of the page should not matter in this case; HTML entity references always refer to Unicode code points.
PHP has a function htmlentities for this purpose, but it appears that you will need workarounds for handling numeric entities. This json_encode hack is fairly obscure, but is probably programmatically the simplest.
echo preg_replace('/\\\\u([0-9a-f]{4})/', '&#x$1;',
preg_replace('^/"(.*)"$/', '$1', json_encode($s)));
This leverages the fact that json_encode will coincidentally do the conversion for you; the rest is all mechanics. (I guess that's PHP for you.)
IDEone demo
Your "bonus question" isn't really a question, but of course, that's how it works; raw bytes in the range 128-255 are only rarely valid UTF-8 sequences, so unless what you have on the page is valid UTF-8, you are likely to get the "invalid character" replacement glyph for those bytes.
For the record, the first two Chinese Han glyphs in your text in UTF-8 would display as æ ¼æ‹‰ if mistakenly displayed in Windows code page 1252 (what you, and oftentimes Microsoft, carelessly refer to as "ANSI") -- if you have those bytes on the page then forcing the browser to display it in UTF-8 should actually work as a workaround as well.
For additional background I recommend @deceze's What Every Programmer Absolutely, Positively Needs to Know About Encodings and Character Sets to Work With Text.
