Merge branch 'master' of https://github.com/TheDrawingCoder-Gamer/nheko into TheDrawingCoder-Gamer-master

.gitignore

@@ -7,6 +7,7 @@ cscope*
 /.exrc
 .gdb_history
 .hunter
+resources/complete-emoji.txt 
 
 # GTAGS
 GTAGS

resources/extra_emoji.txt

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

resources/provider-header.txt

@@ -0,0 +1,10 @@
+// SPDX-FileCopyrightText: 2021 Nheko Contributors
+// SPDX-FileCopyrightText: 2022 Nheko Contributors
+//
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+#include "emoji/Provider.h"
+
+using namespace emoji;
+
+

resources/qml/emoji/EmojiPicker.qml

@@ -146,7 +146,7 @@ Menu {
                     width: 48
                     height: 48
                     hoverEnabled: true
-                    ToolTip.text: model.shortName
+                    ToolTip.text: model.toolTip
                     ToolTip.visible: hovered
                     // TODO: maybe add favorites at some point?
                     onClicked: {

resources/shortcodes.txt

@@ -0,0 +1,55 @@
+1F62D:sob
+1F631:scream
+1F613:sweat 
+1F605:sweat_smile
+1F604:smile 
+1F923:rofl 
+1F60B:yum
+1F607:innocent
+1F917:hugging_face
+26A1:zap
+203C FE0F:bangbang
+2049 FE0F:interrobang
+2753:question 
+2757:exclamation 
+2754:white_question
+2755:white_exclamation
+1F520:big_abcd
+1F521:abcd 
+1F522:1234
+1F523:symbols 
+1F524:abc
+1F60D:heart_eyes
+1F927:cry_smile
+1F608:smiling_imp 
+1F47F:imp
+1F261:rage
+1F624:triumph
+1F63A:smiley_cat
+1F638:smile_cat 
+1F639:joy_cat
+1F63B:heart_eyes_cat
+1F63C:smirk_cat
+1F640:scream_cat
+1F4AF:100
+1F44B:wave
+1F448:point_left
+1F449:point_right
+1F446:point_up_2
+1F447:point_down
+261D FE0F:point_up 
+1FAF5:point_you
+1F937:shrug
+0023 FE0F 20E3:hash 
+002A FE0F 20E3:asterisk 
+0030 FE0F 20E3:zero 
+0031 FE0F 20E3:one 
+0032 FE0F 20E3:two
+0033 FE0F 20E3:three
+0034 FE0F 20E3:four 
+0035 FE0F 20E3:five
+0036 FE0F 20E3:six
+0037 FE0F 20E3:seven
+0038 FE0F 20E3:eight
+0039 FE0F 20E3:nine
+1F51F:ten

scripts/codegen.sh

@@ -0,0 +1,7 @@
+#!/bin/bash
+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
+scripts/emoji_codegen.py resources/complete-emoji.txt resources/shortcodes.txt >> src/emoji/Provider.cpp
+cd - > /dev/null

scripts/emoji_codegen.py

@@ -2,14 +2,15 @@
 
 import sys
 import re
-
+from unidecode import unidecode
 from jinja2 import Template
 
 
 class Emoji(object):
-    def __init__(self, code, shortname):
+    def __init__(self, code, shortname, unicodename):
         self.code = ''.join(['\\U'+c.rjust(8, '0') for c in code.strip().split(' ')])
         self.shortname = shortname
+        self.unicodename = unicodename
 
 def generate_qml_list(**kwargs):
     tmpl = Template('''
@@ -17,20 +18,20 @@ const QVector<Emoji> emoji::Provider::emoji = {
     {%- for c in kwargs.items() %}
     // {{ c[0].capitalize() }}
     {%- for e in c[1] %}
-    Emoji{QStringLiteral(u"{{ e.code }}"), QStringLiteral(u"{{ e.shortname }}"), emoji::Emoji::Category::{{ c[0].capitalize() }}},
+    Emoji{QStringLiteral(u"{{ e.code }}"), QStringLiteral(u"{{ e.shortname }}"), QStringLiteral(u"{{ e.unicodename }}"), emoji::Emoji::Category::{{ c[0].capitalize() }}},
     {%- endfor %}
     {%- endfor %}
 };
     ''')
     d = dict(kwargs=kwargs)
     print(tmpl.render(d))
-
 if __name__ == '__main__':
-    if len(sys.argv) < 2:
+    if len(sys.argv) < 3:
-        print('usage: emoji_codegen.py /path/to/emoji-test.txt')
+        print('usage: emoji_codegen.py /path/to/emoji-test.txt /path/to/shortcodes.txt')
         sys.exit(1)
 
     filename = sys.argv[1]
+    shortcodefilename = sys.argv[2]
 
     people = []
     nature = []
@@ -50,9 +51,14 @@ if __name__ == '__main__':
         'Activities': activity,
         'Objects': objects,
         'Symbols': symbols,
-        'Flags': flags
+        'Flags': flags,
+        'Component': symbols
     }
-
+    shortcodeDict = {} 
+    # for my sanity - this strips newlines
+    for line in open(shortcodefilename, 'r', encoding="utf8"): 
+        longname, shortname = line.strip().split(':')
+        shortcodeDict[longname] = shortname
     current_category = ''
     for line in open(filename, 'r', encoding="utf8"):
         if line.startswith('# group:'):
@@ -68,16 +74,65 @@ if __name__ == '__main__':
         code, qualification, charAndName = segments
 
         # skip unqualified versions of same unicode
-        if qualification == 'unqualified':
+        if qualification != 'fully-qualified':
             continue
         
-        if qualification == 'component':
-            continue
 
         char, name = re.match(r'^(\S+) E\d+\.\d+ (.*)$', charAndName).groups()
-
+        shortname = name
-        categories[current_category].append(Emoji(code, name))
+        # until skin tone is handled, keep them around
+        # discard skin tone variants for sanity
+        # __contains__ is so stupid i hate prototype languages
+        # if name.__contains__("skin tone") and qualification != 'component': 
+        #    continue
+        # if qualification == 'component' and not name.__contains__("skin tone"): 
+        #    continue
+        #TODO: Handle skintone modifiers in a sane way
+        basicallyTheSame = False
+        if code in shortcodeDict: 
+            shortname = shortcodeDict[code]
+        else:
+            shortname = shortname.lower()
+            if shortname.endswith(' (blood type)'): 
+                shortname = shortname[:-13]
+            if shortname.endswith(': red hair'): 
+                shortname = "red_haired_" + shortname[:-10]
+            if shortname.endswith(': curly hair'): 
+                shortname = "curly_haired_" + shortname[:-12]
+            if shortname.endswith(': white hair'): 
+                shortname = "white_haried_" + shortname[:-12]
+            if shortname.endswith(': bald'): 
+                shortname = "bald_" + shortname[:-6]
+            if shortname.endswith(': beard'): 
+                shortname = "bearded_" + shortname[:-7]
+            if shortname.endswith(' face'): 
+                shortname = shortname[:-5]
+            if shortname.endswith(' button'): 
+                shortname = shortname[:-7] 
+            if shortname.endswith(' banknote'): 
+                shortname = shortname[:-9]
+                
+            # FIXME: Is there a better way to do this?
+            matchobj = re.match(r'^flag: (.*)$', shortname) 
+            if shortname.startswith("flag: "): 
+                country = shortname[5:]
+                shortname = country + " flag"
+            shortname = shortname.replace("u.s.", "us")
+            shortname = shortname.replace("&", "and")
+            
+            if shortname == name.lower(): 
+                basicallyTheSame = True
+
+            shortname = shortname.replace("-", "_")
+            shortname = re.sub(r'\W', '_', shortname)
+            shortname, = re.match(r'^_*(.+)_*$', shortname).groups()
+            shortname = re.sub(r'_{2,}', '_', shortname) 
+            shortname = unidecode(shortname)
+        # if basicallyTheSame: 
+        #    shortname = ""
+        categories[current_category].append(Emoji(code, shortname, name))
 
     # Use xclip to pipe the output to clipboard.
     # e.g ./codegen.py emoji.json | xclip -sel clip
+    # alternatively - delete the var from src/emoji/Provider.cpp, and do ./codegen.py emojis shortcodes >> src/emoji/Provider.cpp
     generate_qml_list(people=people, nature=nature, food=food, activity=activity, travel=travel, objects=objects, symbols=symbols, flags=flags)

scripts/update_emoji.md

@@ -2,6 +2,8 @@
 
 1. Get the latest emoji-test.txt from here: https://unicode.org/Public/emoji/
 2. Overwrite the existing resources/emoji-test.txt with the new one
-3. Run `./scripts/emoji_codegen.py resources/emoji-test.txt` and replace the current tail of src/emoji/Provider.cpp with the new output
+3. Run `./scripts/emoji_codegen.py resources/emoji-test.txt resources/shortcodes.txt` and replace the current tail of src/emoji/Provider.cpp with the new output
 4. `make lint`
 5. Compile and test
+
+

src/emoji/EmojiModel.cpp

@@ -34,6 +34,7 @@ EmojiModel::roleNames() const
         roles                                                   = QAbstractListModel::roleNames();
         roles[static_cast<int>(EmojiModel::Roles::Unicode)]     = QByteArrayLiteral("unicode");
         roles[static_cast<int>(EmojiModel::Roles::ShortName)]   = QByteArrayLiteral("shortName");
+        roles[static_cast<int>(EmojiModel::Roles::UnicodeName)] = QByteArrayLiteral("unicodeName");
         roles[static_cast<int>(EmojiModel::Roles::Category)]    = QByteArrayLiteral("category");
         roles[static_cast<int>(EmojiModel::Roles::Emoji)]       = QByteArrayLiteral("emoji");
     }
@@ -58,10 +59,12 @@ EmojiModel::data(const QModelIndex &index, int role) const
             return Provider::emoji[index.row()].unicode;
 
         case Qt::ToolTipRole:
+        case CompletionModel::SearchRole2:
+        case static_cast<int>(EmojiModel::Roles::UnicodeName):
+            return Provider::emoji[index.row()].unicodeName;
         case CompletionModel::SearchRole:
         case static_cast<int>(EmojiModel::Roles::ShortName):
             return Provider::emoji[index.row()].shortName;
-
         case static_cast<int>(EmojiModel::Roles::Category):
             return QVariant::fromValue(Provider::emoji[index.row()].category);
 

src/emoji/EmojiModel.h

@@ -26,6 +26,7 @@ public:
         Unicode = Qt::UserRole, // unicode of emoji
         Category,               // category of emoji
         ShortName,              // shortext of the emoji
+        UnicodeName,            // true unicode name of the emoji
         Emoji,                  // Contains everything from the Emoji
     };
 

src/emoji/Provider.h

@@ -35,11 +35,13 @@ public:
 
     Q_PROPERTY(const QString &unicode MEMBER unicode)
     Q_PROPERTY(const QString &shortName MEMBER shortName)
+    Q_PROPERTY(const QString &unicodeName MEMBER unicodeName)
     Q_PROPERTY(emoji::Emoji::Category category MEMBER category)
 
 public:
     QString unicode;
     QString shortName;
+    QString unicodeName;
     Category category;
 };