Use Janome and NEologd instead of pykakasi

Related to #32
Close #31

Dockerfile

@@ -19,6 +19,10 @@ RUN python3 -m pip install --no-cache-dir --upgrade pip && \
     python3 -m pip install --no-cache-dir -r ./requirements.txt && \
     python3 -m pip check
 
+RUN wget -q https://github.com/peaceiris/emoji-ime-dictionary/releases/download/v2.2.1/Janome-0.4.1-neologd-20200910.tar.gz && \
+    python3 -m pip install --no-cache-dir --no-compile Janome-0.4.1-neologd-20200910.tar.gz && \
+    rm Janome-0.4.1-neologd-20200910.tar.gz
+
 RUN wget -q 'https://raw.githubusercontent.com/yagays/emoji-ja/20190726/data/emoji_ja.json' \
       -O /root/emoji_ja.json
 

Makefile

@@ -1,6 +1,6 @@
 .PHONY: build
 build:
-	docker-compose run --rm -T dev python main.py
+	COMPOSE_DOCKER_CLI_BUILD=1 docker-compose run --rm -T dev python main.py
 
 .PHONY: test
 test:

main.py

@@ -1,8 +1,9 @@
-import urllib.request
 import json
+from janome.tokenizer import Tokenizer
+from pykakasi import kakasi
+import re
 
-from pykakasi import kakasi, wakati
-
+t = Tokenizer()
 
 kakasi = kakasi()
 kakasi.setMode("J","H")
@@ -11,13 +12,24 @@
 conv_k2h = kakasi.getConverter()
 
 
-def hiraganafy(keyword):
+def hiraganafy_v1(keyword):
     k = keyword.upper()
     k = conv_j2h.do(k)
     k = conv_k2h.do(k)
     return k
 
 
+def hiraganafy(keyword):
+    katakana = ''
+    for token in t.tokenize(keyword.upper()):
+        katakana += token.surface if token.reading == '*' else token.reading
+    hiragana = conv_k2h.do(katakana)
+    hiragana_v1 = hiraganafy_v1(keyword)
+    if hiragana != hiragana_v1:
+        print(f'| {keyword} | {hiragana} | {hiragana_v1} |')
+    return hiragana
+
+
 def add_word_to_dict(emoji, keyword, emoji_dict):
     valid_keyword = keyword.replace('ゔ', 'う゛')
     word = f':{valid_keyword}\t{emoji}\t記号\t'
@@ -46,14 +58,14 @@ def create_emoji_dict(self) -> None:
             for k in self.emoji_json[emoji]['keywords']:
                 if k.isalpha() is False:
                     continue
-                k = hiraganafy(k)
-                add_word_to_dict(emoji, k, self.emoji_dict)
+                hiragana = hiraganafy(k)
+                add_word_to_dict(emoji, hiragana, self.emoji_dict)
 
             k = self.emoji_json[emoji]['short_name']
             if k.isalpha() is False:
                 continue
-            k = hiraganafy(k)
-            add_word_to_dict(emoji, k, self.emoji_dict)
+            hiragana = hiraganafy(k)
+            add_word_to_dict(emoji, hiragana, self.emoji_dict)
 
         self.emoji_dict.sort()
         self.emoji_dict = '\n'.join(self.emoji_dict)

main_test.py

@@ -15,22 +15,12 @@ def test_hiraganafy(self):
             {'in': '営マーク', 'out': 'えいまーく'},
             {'in': '営業中', 'out': 'えいぎょうちゅう'},
             {'in': '営業日', 'out': 'えいぎょうび'},
-        ]
-        for t in test_cases:
-            self.assertEqual(main.hiraganafy(t['in']), t['out'])
-
-    def test_hiraganafy_todo(self):
-        test_cases = [
-            {'in': 'しかめ面', 'out': 'しかめっつら'},
-            {'in': 'しかめ面の人', 'out': 'しかめつらのひと'},
             {'in': 'バルカン人', 'out': 'ばるかんじん'},
-            {'in': 'フェイスマッサージ中の人', 'out': 'ふぇいすまっさーじちゅうのひと'},
             {'in': '水球をする人', 'out': 'すいきゅうをするひと'},
             {'in': '人のシルエット', 'out': 'ひとのしるえっと'},
             {'in': '話す人のシルエット', 'out': 'はなすひとのしるえっと'},
             {'in': '2人のシルエット', 'out': 'ふたりのしるえっと'},
             {'in': '六芒星', 'out': 'ろくぼうせい'},
-            {'in': '六角星', 'out': 'ろっかくせい'},
             {'in': '波乗り', 'out': 'なみのり'},
             {'in': 'きらきら星', 'out': 'きらきらぼし'},
             {'in': 'くす玉', 'out': 'くすだま'},
@@ -40,6 +30,16 @@ def test_hiraganafy_todo(self):
             {'in': '介助犬', 'out': 'かいじょけん'},
             {'in': '筋トレ', 'out': 'きんとれ'},
         ]
+        for t in test_cases:
+            self.assertEqual(main.hiraganafy(t['in']), t['out'])
+
+    def test_hiraganafy_todo(self):
+        test_cases = [
+            {'in': 'しかめ面', 'out': 'しかめっつら'},
+            {'in': 'しかめ面の人', 'out': 'しかめつらのひと'},
+            {'in': 'フェイスマッサージ中の人', 'out': 'ふぇいすまっさーじちゅうのひと'},
+            {'in': '六角星', 'out': 'ろっかくせい'},
+        ]
         for t in test_cases:
             self.assertNotEqual(main.hiraganafy(t['in']), t['out'])