Merge branch 'jp_pronunciation' into develop

lute/models/setting.py

@@ -150,6 +150,7 @@ def load():
             "backup_dir": app_config.default_user_backup_path,
             "backup_count": 5,
             "mecab_path": None,
+            "japanese_reading": "katakana",
             "current_theme": "-",
             "custom_styles": "/* Custom css to modify Lute's appearance. */",
             "show_highlights": True,

lute/parse/mecab_parser.py

@@ -16,7 +16,9 @@
 import re
 from typing import List
 from natto import MeCab
+import jaconv
 from lute.parse.base import ParsedToken, AbstractParser
+from lute.models.setting import UserSetting
 
 
 class JapaneseParser(AbstractParser):
@@ -136,4 +138,12 @@ def get_reading(self, text: str):
         ret = "".join(readings).strip()
         if ret in ("", text):
             return None
-        return ret
+
+        jp_reading_setting = UserSetting.get_value("japanese_reading")
+        if jp_reading_setting == "katakana":
+            return ret
+        if jp_reading_setting == "hiragana":
+            return jaconv.kata2hira(ret)
+        if jp_reading_setting == "alphabet":
+            return jaconv.kata2alphabet(ret)
+        raise RuntimeError(f"Bad reading type {jp_reading_setting}")

lute/settings/routes.py

@@ -47,6 +47,12 @@ class UserSettingsForm(FlaskForm):
     show_highlights = BooleanField("Highlight terms by status")
 
     mecab_path = StringField("MECAB_PATH environment variable")
+    reading_choices = [
+        ("katakana", "Katakana"),
+        ("hiragana", "Hiragana"),
+        ("alphabet", "Romaji"),
+    ]
+    japanese_reading = SelectField("Pronunciation characters", choices=reading_choices)
 
     def validate_backup_dir(self, field):
         "Field must be set if enabled."

lute/templates/settings/form.html

@@ -85,21 +85,11 @@ <h2>Appearance</h2>
 
     <tr>
       <td>
-        <h2>MeCab (Japanese only)</h2>
+        <h2>Japanese</h2>
       </td>
       <td />
     </tr>
 
-    <tr>
-      <td>
-        <button onclick="test_mecab_setup(); return false;">
-          Test my MeCab configuration
-        </button>
-      </td>
-      <td>
-        <div id="parse_result"></div>
-      </td>
-    </tr>
     <tr>
       <td>
         {{ form.mecab_path.label }}
@@ -127,6 +117,20 @@ <h2>MeCab (Japanese only)</h2>
       </td>
       <td>{{ form.mecab_path(class="form-control") }}</td>
     </tr>
+    <tr>
+      <td>
+      </td>
+      <td>
+        <button onclick="test_mecab_setup(); return false;">
+          Test my MeCab configuration
+        </button>
+        <div style="margin-top: 5px" id="parse_result"></div>
+      </td>
+    </tr>
+    <tr>
+      <td>{{ form.japanese_reading.label }}</td>
+      <td>{{ form.japanese_reading }}</td>
+    </tr>
 
   </table>
 

pyproject.toml

@@ -23,6 +23,7 @@ dependencies = [
   "Flask-SQLAlchemy>=3.1.1,<4",
   "Flask-WTF>=1.2.1,<2",
   "natto-py>=1.0.1,<2",
+  "jaconv>=0.3.4,<1",
   "platformdirs>=3.10.0,<4",
   "requests>=2.31.0,<3",
   "beautifulsoup4>=4.12.2,<5",

requirements.txt

@@ -28,6 +28,7 @@ iniconfig==2.0.0
 invoke==2.2.0
 isort==5.12.0
 itsdangerous==2.1.2
+jaconv==0.3.4
 Jinja2==3.1.2
 lazy-object-proxy==1.9.0
 Mako==1.2.4

tests/unit/parse/test_JapaneseParser.py

@@ -4,6 +4,8 @@
 
 from lute.parse.mecab_parser import JapaneseParser
 from lute.models.term import Term
+from lute.models.setting import UserSetting
+from lute.db import db
 from lute.parse.base import ParsedToken
 
 
@@ -74,3 +76,17 @@ def test_readings():
 
     for c in cases:
         assert p.get_reading(c[0]) == c[1], c[0]
+
+
+def test_reading_setting(app_context):
+    "Return reading matching user setting."
+    cases = {
+        "katakana": "ツヨイ",
+        "hiragana": "つよい",
+        "alphabet": "tsuyoi",
+    }
+    p = JapaneseParser()
+    for k, v in cases.items():
+        UserSetting.set_value("japanese_reading", k)
+        db.session.commit()
+        assert p.get_reading("強い") == v, k