Bug 503451

Created attachment 180717 [details]
screen shots

SUMMARY
KCharSelect seems to have a glitch when faced with characters not in a given font.

It has another problem, that I have not encountered before, where if one selects "All" from the dropdown menu that lets you select which Unicode block (e.g. "European Scripts" etc it locks up and won't even quit with clicking on the close X or when "close all windows" are chosen by right clicking on its icon in the task manager.  It will quit w/ "killall..." however.

I think KCharSelect used to only show those characters/glyphs included in a given font.   Now it seems to show fonts from other characters.  In most cases, only when the character is not present, but in one example (using EmojiOne either regular or color) it will substitute a character when there is not one defined in the font, but shows a blank/gap when there is in fact an actually defined character.  In this example, I noticed that even in the text display are (e.g. Character: ⚕ U+2695 / Name: STAFF OF AESCULAPIUS) when I selected the blank spot for the character it displayed a colored/emoji style version but NOT THE ONE FROM THE CHOSEN FONT (I think it pulled the character that belongs to Noto Emoji, but am not 100% certain on that).

On my system, it seems to pull replacement characters from other fonts when not available, which isn't desirable default behavior as it can result in me selecting and pasting in a character that is neither in the font I am using and quite possible not present on other's systems.

I cannot access all the defined Unicode characters in a font as there does not seem to be any way to access the Mathematical Alphanumeric Symbols (https://unicode.org/charts/nameslist/n_1D400.html) block, which have been part of the Unicode standard for over twenty years.  

This *should* be one of the character range blocks presented in the first drop-down menu (e.g. "European scripts..." etc. etc.) menu.  Q.v.  https://en.wikipedia.org/wiki/Mathematical_Alphanumeric_Symbols, https://unicode.org/charts/PDF/U1D400.pdf).

E.g. the character 𝑥 is often used as an algebraic unknown.  If I paste it into the KCharSelect find I get details, e.g. that it is NON-BMP-CHARACTER-1D465, 
Various Useful Representations
UTF-8: 0xF0 0x9D 0x91 0xA5
UTF-16: 0xD835 0xDC65
C octal escaped UTF-8: \360\235\221\245
XML decimal entity: 𝑥

Etc. but no way to even browse for, or search for,  the character even if a font (e.g. I used DejaVu Math TeX Gyre here, but  IBM Plex Math, Noto Sans Math, Quivira, and  GNU Free Serif)  contains the range.  As mentioned, it is also problematic, is that if I switch to a font that I know does NOT include this, it still shows up!

I know in the past I have had a great deal of difficulty reading PDFs that someone had pasted in symbols using one of those non-Unicode "symbol" fonts which resulted in a non-legible document on my end.  I would wish to avoid being the cause of such problems for others by making sure that any special characters were either in the specified font, or that I at least knew I as using a replacement font!

Rather, an option to toggle "show replacement characters when not present in selected font" would be very helpful and avoid confusion (and selection of a character by accident which isn't actually present in a given font).

For those who have the LastResort or Uniding font installed, a gray colored replacement character would be a reasonable option to indicate when a defined character isn't present in a font.  Another possible, helpful and appreciated, option would be to have a colored (to make it obvious) version of Unifont (if installed, and perhaps having the bitmap and TTF GNU GNU Unifont as a recommended package , as well as the other fonts, e.g. Unifont/Unidings/LastResort/Nerdfont Symbols as "suggested" packages, based on what various distributions had at hand in their repos, might be wise/helpful)

For future consideration, it would be very helpful to add an option to preview a given character in other fonts (all/default replacement font/user selected).  I have used the GNOME Font Manager in the past, esp. to access formatted (e.g. italic, script, bold, serif/sanserif) Latin and Greek characters that are in a Unicode block typically used for variables/constants/functions in mathematic/logic notations that I was not able to locate using KCharSelect.  For example, the characters used to say:  𝒻(𝒙) = 𝓎 I could not figure out how to access with KCharSelect.

I have noticed that some PUA characters, specifically some from NerdFont use,  act similarly, while some, if I paste them in and search for them, I can browse the other characters in the PUA space.  It seems to depend on the actual code point, as those which use a more or less conventional code point, when I deleted them from the search area, let me browse the PUA, but more unconventional mappings.  E.g.  (U+EAC9) shows up as Tux with Devicon, but not other fonts.    (U+F17C), as well    (U+F31A)  shows up as Tux, regardless of which font I have selected (including those which do not have any PUA characters)  .  If I cut-and-paste the character into the search area, regardless of which font is selected, and then delete it, I can browse the various PUA characters (again, even if I have selected a font that barely contains the basic Latin character set or one of the TTF versions of a historic VGA font).  It seems that in creation of the Nerd fonts that they have mapped in their sources that they have assigned duplicate glyphs--or at least visually indistinct--to different code points.  Thus, there may be a multiple identically appearing characters with different assigned Unicode PUA values (which isn't exactly how I would have done it).  E.g. if you go to their "cheat sheet" and search for "git" you will end up with many different (Unicode values), but identical (graphically) options for the GIT logo. 

For those PUA glyphs where there is a degree of consistency (e.g. FontAwesome, NerdFonts,  Catrinity agree, and NerdFont and Catrinity names are the same/similar) it would be super helpful to be able to display the name of the character and be able to search for it.  I know these are not nearly as standard as the Unicode (and I would prioritize getting all of the Unicode, esp. the math characters, ahead of the PUA 'gentlemen's agreement').  

Having a link between a character's description and the associated Wikipedia entry (in the user's default language) and/or Unicode website might also be helpful and fairly easy to implement.

Finally, there is a growing convention about many of the PUA code points.  Perhaps including some sort of description, or ability to look up PUA font on the Nerdfont "cheat sheet" (maybe something that users should select as an option or base it on having a Nerdfont installed)

Thanks for looking into this.  I use this small app nearly every day!

SOFTWARE/OS VERSIONS
KCharSelect: 25.04.0
KDE Frameworks: 6.13.0
Qt: Using 6.9.0 and built against 6.9.0
openSUSE Tumbleweed (Xcb)
Build ABI: x86_64-little_endian-lp64
Kernel: linux 6.14.2-1-default