Add standard emoji codes

[LucyMcGowan]

I think it'd be lovely if we could incorporate these emoji codes - used by Slack, GitHub, etc 🎉 👯

@romainfrancois: Here is their GitHub - I haven't poked around much, but can if you think this is worthwhile.

[romainfrancois]

Cool thanks for logging it here. I've had a quick look and I only found how to match the codes with pictures, but I need to :bowtie: this to their actual character representations, e.g.

> slice( emo::jis , 1:5  )
# A tibble: 5 x 17
     id        emoji                           name         category   subcategory  keywords skin_tone     runes
  <dbl>        <chr>                          <chr>            <chr>         <chr>    <list>     <chr>    <list>
1     1 "\U0001f600"                  grinning face Smileys & People face-positive <chr [2]>      <NA> <chr [1]>
2     2 "\U0001f601" beaming face with smiling eyes Smileys & People face-positive <chr [4]>      <NA> <chr [1]>
3     3 "\U0001f602"         face with tears of joy Smileys & People face-positive <chr [4]>      <NA> <chr [1]>
4     4 "\U0001f923"  rolling on the floor laughing Smileys & People face-positive <chr [4]>      <NA> <chr [1]>
5     5 "\U0001f603"    grinning face with big eyes Smileys & People face-positive <chr [4]>      <NA> <chr [1]>
# ... with 9 more variables: nrunes <int>, apple <lgl>, google <lgl>, twitter <lgl>, one <lgl>, facebook <lgl>,
#   messenger <lgl>, samsung <lgl>, windows <lgl>

[LucyMcGowan]

I think there are a couple of places that have mapped them (for example, here is python's and here is C's), so maybe we can use that?

[romainfrancois]

Perfect. Should be easy to use one of those. Not sure if this goes in a list column in jis or in a separate simple named chr vector.

The hesitation is bc some have multiple names, eg iirc :poop:

I like the tibble but it might be more practical

[romainfrancois]

Good news from looking at the python link above, they use this to transform some of the names from the unicode names (which we already have in pull( emo::jis, name )

> sample( dplyr::pull( emo::jis, name ), 10 )
 [1] "tear-off calendar"                      "woman juggling: medium-light skin tone" "purse"                                 
 [4] "information"                            "woman climbing: medium skin tone"       "baby angel"                            
 [7] "man cartwheeling: medium skin tone"     "poultry leg"                            "South Georgia & South Sandwich Islands"
[10] "input numbers"                         

so I guess to get started this will be a str_replace_all exercise in data-raw/jis.R

Then, they use the aliases, defined here

EMOJI_ALIAS_UNICODE = dict(EMOJI_UNICODE.items(), **{
    u':admission_tickets:': u'\U0001F39F',
    u':aerial_tramway:': u'\U0001F6A1',
    u':airplane:': u'\U00002708',
    u':airplane_arriving:': u'\U0001F6EC',
    u':airplane_departure:': u'\U0001F6EB',

I guess we can parse this somehow, but this also looks generated automatically ...

[romainfrancois]

ah, it looks like it all goes down to this repo: https://github.com/github/gemoji and particularly this file which as a bonus also has information about the version of unicode when the emoji was introduced:

... 
{
    "emoji": "😀"
  , "description": "grinning face"
  , "category": "People"
  , "aliases": [
      "grinning"
    ]
  , "tags": [
      "smile"
    , "happy"
    ]
  , "unicode_version": "6.1"
  , "ios_version": "6.0"
  } 
... 

[romainfrancois]

We now have emo::ji_alias vector (which might replace emo:::emoji_name ) so :

> cat( emo::ji_alias["grin"] )
😁

I'll simplify ji_glue to use this

[romainfrancois]

@LucyMcGowan For now I've put this in ji_replace but not 💍 with the name.

> ji_replace( ":ant: != :santa: " )
🐜 != 🎅 

[romainfrancois]

it probably should hint (did you mean ...) when it does not match

[LucyMcGowan]

Could it just be emo::ji_glue()? The other thing I was 🤔 is whether it is odd that ordinarily glue uses {parameter} but we use :parameter:, the latter I think makes sense because that's how all other places do (i.e.: Slack, GitHub, etc), but worth a thought.

[romainfrancois]

Makes sense for this to be ji_glue, btw it can't actually use glue i think as the begin and end are the same.

I kind of want to keem what i have now as i can do either :cat,face: or :face,cat: ... but i guess for most people, esp who know slack github ... it makes more sense to stick to exact matches.

To be truly great, we need rstudio to show emoji completion. @kevinushey shared some hints about where completion is implemented in rstudio, this might be my excuse to be curious about rstudio internals ...

I think my plan is:

• use exact matches based on ji_alias and/or sample(1) from exact keywords matches in ji and ji_glue
• move and improve some of the quirky syntax i have in jitsu into ji_find