diff options
| author | emkael <emkael@tlen.pl> | 2021-04-02 03:37:29 +0200 |
|---|---|---|
| committer | emkael <emkael@tlen.pl> | 2021-04-02 03:37:29 +0200 |
| commit | bb8d5d0520e078f157448a7d3b4ce196f447a627 (patch) | |
| tree | 5c914af2d96513ef75d28f06e44130c7e6ee8ce6 | |
| parent | 24bd87bcaf315c761f85dc898fcaf9a715b630c4 (diff) | |
Filtering out Unicode retard-speak
| -rw-r--r-- | bin/emoji-list.py | 8 | ||||
| -rw-r--r-- | config/emoji.json | bin | 77700 -> 96131 bytes |
2 files changed, 8 insertions, 0 deletions
diff --git a/bin/emoji-list.py b/bin/emoji-list.py index 31ba4b1..823347d 100644 --- a/bin/emoji-list.py +++ b/bin/emoji-list.py @@ -1,4 +1,5 @@ import json, os, urllib +import xml.etree.ElementTree as ET emoji_list = json.load(urllib.urlopen( 'https://raw.githubusercontent.com/iamcal/emoji-data/master/emoji.json' @@ -9,6 +10,13 @@ for emoji in emoji_list: name = emoji['name'].replace(' ', '_') if emoji['name'] else emoji['short_name'].upper().replace('-', '_') dictionary[character] = ' [%s] ' % (name) +retardspeak_map = ET.fromstring(urllib.urlopen( + 'http://slothsoft.net/getResource.php/slothsoft/unicode-mapper' +).read()) +for letter in retardspeak_map.findall('.//letter'): + if letter.attrib['target'] != letter.attrib['source']: + dictionary[letter.attrib['target']] = letter.attrib['source'] + json.dump(dictionary, file(os.path.join( os.path.dirname(os.path.realpath(__file__)), '../config/emoji.json' diff --git a/config/emoji.json b/config/emoji.json Binary files differindex 161831d..f3992c2 100644 --- a/config/emoji.json +++ b/config/emoji.json |