Properly add regional indicators, with ZWNJ

resources/extra_emoji.txt

@@ -1,27 +1,27 @@
 # group: Symbols 
-1F1E6 ; fully-qualified # 🇦 E6.0 regional indicator a 
+200C 1F1E6 200C ; fully-qualified # ‌🇦‌ E6.0 regional indicator a
-1F1E7 ; fully-qualified # 🇧 E6.0 regional indicator b 
+200C 1F1E7 200C ; fully-qualified # ‌🇧‌ E6.0 regional indicator b
-1F1E8 ; fully-qualified # 🇨 E6.0 regional indicator c
+200C 1F1E8 200C ; fully-qualified # ‌🇨‌ E6.0 regional indicator c
-1F1E9 ; fully-qualified # 🇩 E6.0 regional indicator d 
+200C 1F1E9 200C ; fully-qualified # ‌🇩‌ E6.0 regional indicator d
-1F1EA ; fully-qualified # 🇪 E6.0 regional indicator e 
+200C 1F1EA 200C ; fully-qualified # ‌🇪‌ E6.0 regional indicator e
-1F1EB ; fully-qualified # 🇫 E6.0 regional indicator f 
+200C 1F1EB 200C ; fully-qualified # ‌🇫‌ E6.0 regional indicator f
-1F1EC ; fully-qualified # 🇬 E6.0 regional indicator g 
+200C 1F1EC 200C ; fully-qualified # ‌🇬‌ E6.0 regional indicator g
-1F1ED ; fully-qualified # 🇭 E6.0 regional indicator h 
+200C 1F1ED 200C ; fully-qualified # ‌🇭‌ E6.0 regional indicator h
-1F1EE ; fully-qualified # 🇮 E6.0 regional indicator i
+200C 1F1EE 200C ; fully-qualified # ‌🇮‌ E6.0 regional indicator i
-1F1EF ; fully-qualified # 🇯 E6.0 regional indicator j 
+200C 1F1EF 200C ; fully-qualified # ‌🇯‌ E6.0 regional indicator j
-1F1F0 ; fully-qualified # 🇰 E6.0 regional indicator k 
+200C 1F1F0 200C ; fully-qualified # ‌🇰‌ E6.0 regional indicator k
-1F1F1 ; fully-qualified # 🇱 E6.0 regional indicator l 
+200C 1F1F1 200C ; fully-qualified # ‌🇱‌ E6.0 regional indicator l
-1F1F2 ; fully-qualified # 🇲 E6.0 regional indicator m
+200C 1F1F2 200C ; fully-qualified # ‌🇲‌ E6.0 regional indicator m
-1F1F3 ; fully-qualified # 🇳 E6.0 regional indicator n
+200C 1F1F3 200C ; fully-qualified # ‌🇳‌ E6.0 regional indicator n
-1F1F4 ; fully-qualified # 🇴 E6.0 regional indicator o  
+200C 1F1F4 200C ; fully-qualified # ‌🇴‌ E6.0 regional indicator o
-1F1F5 ; fully-qualified # 🇵 E6.0 regional indicator p 
+200C 1F1F5 200C ; fully-qualified # ‌🇵‌ E6.0 regional indicator p
-1F1F6 ; fully-qualified # 🇶 E6.0 regional indicator q 
+200C 1F1F6 200C ; fully-qualified # ‌🇶‌ E6.0 regional indicator q
-1F1F7 ; fully-qualified # 🇷 E6.0 regional indicator r
+200C 1F1F7 200C ; fully-qualified # ‌🇷‌ E6.0 regional indicator r
-1F1F8 ; fully-qualified # 🇸 E6.0 regional indicator s
+200C 1F1F8 200C ; fully-qualified # ‌🇸‌ E6.0 regional indicator s
-1F1F9 ; fully-qualified # 🇹 E6.0 regional indicator t 
+200C 1F1F9 200C ; fully-qualified # ‌🇹‌ E6.0 regional indicator t
-1F1FA ; fully-qualified # 🇺 E6.0 regional indicator u
+200C 1F1FA 200C ; fully-qualified # ‌🇺‌ E6.0 regional indicator u
-1F1FB ; fully-qualified # 🇻 E6.0 regional indicator v
+200C 1F1FB 200C ; fully-qualified # ‌🇻‌ E6.0 regional indicator v
-1F1FC ; fully-qualified # 🇼 E6.0 regional indicator w
+200C 1F1FC 200C ; fully-qualified # ‌🇼‌ E6.0 regional indicator w
-1F1FD ; fully-qualified # 🇽 E6.0 regional indicator x
+200C 1F1FD 200C ; fully-qualified # ‌🇽‌ E6.0 regional indicator x
-1F1FE ; fully-qualified # 🇾 E6.0 regional indicator y
+200C 1F1FE 200C ; fully-qualified # ‌🇾‌ E6.0 regional indicator y
-1F1FF ; fully-qualified # 🇿 E6.0 regional indicator z
+200C 1F1FF 200C ; fully-qualified # ‌🇿‌ E6.0 regional indicator z

scripts/codegen.sh

@@ -2,6 +2,6 @@
 ROOT=$(realpath "$PWD/$(dirname "$0")/..")
 cd $ROOT
 cat resources/provider-header.txt > src/emoji/Provider.cpp 
-# cat resources/extra_emoji.txt resources/emoji-test.txt > resources/complete-emoji.txt
+cat resources/extra_emoji.txt resources/emoji-test.txt > resources/complete-emoji.txt
-scripts/emoji_codegen.py resources/emoji-test.txt resources/shortcodes.txt >> src/emoji/Provider.cpp
+scripts/emoji_codegen.py resources/complete-emoji.txt resources/shortcodes.txt >> src/emoji/Provider.cpp
 cd - > /dev/null

scripts/emoji_codegen.py

@@ -78,7 +78,7 @@ if __name__ == '__main__':
             continue
 
 
-        char, name = re.match(r'^(\S+) E\d+\.\d+ (.*)$', charAndName.strip()).groups()
+        char, name = re.match(r'^(\S+) E\d+\.\d+ (.*)$', charAndName).groups()
         shortname = name
 
         #TODO: Handle skintone modifiers in a sane way