Unicode emoticons with pdflatex

Here is my solution which is partially based on another post from tex.sx I am afraid I cannot find again:

In a nutshell: Download SVG from a database like twemoji (used by Twitter), convert it and scale it appropriately. Pros: Looks exactly like on Twitter or any other platform you desire. Does not require XeLaTeX. Con: Does not work completely automated.

We make an example of the proposed process by inserting the ? emoji into a TeX document.

1. Find the unicode of your emoji

   Using Google Fu, we find out that the code point for our emoji is U+1F984.

2. Download an SVG file of the emoji

   Be sure to use one consistent data source for all of your tweets. We are referring to twemoji for doing this.

   Download the SVG file from twemoji by adjusting this link: https://raw.githubusercontent.com/twitter/twemoji/master/assets/svg/1f984.svg

3. Convert the SVG file into an EPS file

   There are certain packages for doing so such as includesvg, but I don't want to install inkscape on my machine, so I use an online converter:

   Go to https://convertio.co/de/svg-eps/ and insert the URL from step 2. Download the converted file into your LaTeX figures folder.

4. Typeset the emoji in LaTeX

   This is how I could typeset emojis the best way in LaTeX:

   usepackage{textcomp}  % Required for encoding textbigcircle
   usepackage{scalerel}  % Required for emoji scalerel
   
   def?{scalerel*{includegraphics{figures/1f984.eps}}{textrm{textbigcircle}}}
   

   To use it, just write something like:

   I like ?s, and what's about you?
   

And here is MWE for it: https://overleaf.com/read/tjngsyvrfcmy

PS: I always like to hear your simpler ideas!

I use a setup with 3 files: the list of macros/symbols in UTF8 do-emoji.tex:

% Names below are arbitrary; a macro with this name will be defined.
% After editing the list in the next row, (re)run:    xelatex emoji-from-list
foreach [count=P] M/C in {smilie/?,ghost/?,pumpkin/?}
   {doEmoji{C}{M}{P}}

(I think one should avoid whitespace in the list in the first row.) Then the rendering file emoji-from-list.tex:

documentclass[multi=my,crop]{standalone}
usepackage{pgffor}
usepackage{fontspec}
setmainfont{Symbola}
begin{document}
newcommanddoEmoji[3]{%    character, macroname, page
  begin{my}#1end{my}}
include{do-emoji}
end{document}

(process with xelatex emoji-from-list). Finally, the actual LaTeX file includes:

newcommandincludeEmoji[1]{%   With CM fonts and 1.28, it ascends to the top of the capitals, and descends to the bottom of comma.
  ensuremath{vcenter{hbox{includegraphics[page=#1,height=1.28fontcharhtfont`A]{emoji-from-list}}}}}

{newcommanddoEmoji[3]{%   character, macroname, page
    expandafterxdefcsname #2endcsname{noexpandincludeEmoji{#3}}%
    edefnext{noexpandincludeEmoji{#3}}%         % would need more expansion in the next row otherwise
    expandafterexpandafterexpandafterGnewunicodecharexpandafterexpandafterexpandafter{expandafter#1expandafter}%
       expandafter{next}% % (#1=C)
 }%  For a preamble common for several docs: do nothing if file is not present:
 IfFileExists{do-emoji.tex}{input{do-emoji}}{}}

After this, either use a macro (named in do-emoji.tex), or just use the UTF-8 character. (If you do not need macro names, give the same fake name to all of the characters in the list.) It is easy to auto-generate the file do-emoji.tex.

NOTES:

1. I optimize for usage in math; so I use larger emoji than in other answers, and they are vcenter⸣ed.

2. Above, a macro Gnewunicodechar is used. If you do not plan to use these Unicode chars in your document, just remove the last two lines inside the definition of doEmoji above.

3. I do not know about utf8, but for utf8x, Gnewunicodechar may be defined as this:

   defUNItoNUMz{%        Actually, do not need hex in what follows!
     edefOUT{thecount0}%
   }
   defUNItoNUMcont#1#2{%
     multiplycount0 64relax
     advancecount0 -"80relax
     advancecount0 `#2relax
     #1}
   
   defUNItoNUM#1{%       No attempt is made to detect out-of-range bytes
     count0=`#1%
     ifnum`#1<"C0relax   % In the range a0..bf would give false positives
       letnext=UNItoNUMz
     else
       ifnum`#1<"E0relax
         advancecount0 -"C0relax
         defnext{UNItoNUMcontUNItoNUMz}
       else
         ifnum`#1<"F0relax
           advancecount0 -"E0relax
           defnext{UNItoNUMcont{UNItoNUMcontUNItoNUMz}}
         else
           ifnum`#1<"F8relax
             advancecount0 -"F0relax
             defnext{UNItoNUMcont{UNItoNUMcont{UNItoNUMcontUNItoNUMz}}}
           fi
         fi
       fi
     fi
     next}
   
   newcommandGnewunicodechar[2]{{%   (Since we need global for foreach, localize changes anyway.)  Specific for utf8x
     defgdefUNI##1##2{%
       expandafterletcsname uc@temp@aendcsnameglobalcsname uni@declcharoptendcsname{##1}{document}{##2}}%
     UNItoNUM#1%                  % (#1=C; sets OUT); This is for utf8x; in utf8, may need something else
     gdefUNI{OUT}{#2}%       % finish definition
   }}
   

4. Update: I changed include to (the most robust combination of) input and IfFileExists. The reason: include may create a spurious pagebreak. One should also consider the variant in the first comment below.

Prepare the following file

% smilie.tex
documentclass[preview]{standalone}
usepackage{fontspec}
setmainfont{DejaVu Sans}
begin{document}
?
end{document}

and compile it with XeLaTeX. Then you can use the glyph via the so built PDF file:

documentclass{article}
usepackage[utf8]{inputenc}
usepackage{newunicodechar}
usepackage{graphicx}
newunicodechar{?}{includegraphics{smilie}}

begin{document}
Here is a ?.
end{document}

Probably some tweaking with the borders in the standalone file is necessary.

You can get the height of an uppercase letter in the current font by saying

includegraphics[height=fontcharhtfont`A]{smilie}