Spaces:
Running
Running
Deploy frontend
Browse files- README.md +2 -2
- app.py +26 -20
- requirements.txt +1 -0
README.md
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
---
|
| 2 |
-
title: Parayomi ふりがな
|
| 3 |
-
emoji:
|
| 4 |
colorFrom: green
|
| 5 |
colorTo: red
|
| 6 |
sdk: gradio
|
|
|
|
| 1 |
---
|
| 2 |
+
title: Parayomi — ふりがな予測モデル
|
| 3 |
+
emoji: 📖
|
| 4 |
colorFrom: green
|
| 5 |
colorTo: red
|
| 6 |
sdk: gradio
|
app.py
CHANGED
|
@@ -12,6 +12,7 @@ from __future__ import annotations
|
|
| 12 |
import os
|
| 13 |
|
| 14 |
import gradio as gr
|
|
|
|
| 15 |
from gradio_client import Client
|
| 16 |
|
| 17 |
# ---------------------------------------------------------------------------
|
|
@@ -24,13 +25,14 @@ DEFAULT_TEXT = "深夜の路地は人気が無くて怖い。"
|
|
| 24 |
|
| 25 |
EXAMPLES = [
|
| 26 |
DEFAULT_TEXT,
|
| 27 |
-
"
|
| 28 |
-
"
|
| 29 |
-
"
|
|
|
|
|
|
|
| 30 |
"金の時計を買うために、一生懸命に金を貯めた。",
|
| 31 |
"あちらの方がお見えになった理由は、皆まで言わずとも分かる。",
|
| 32 |
"天気いいし、皆で表に出て遊ぼ?",
|
| 33 |
-
"この表の右上の角に、通し番号を記入してください。",
|
| 34 |
"カブトムシの立派な角に止まった小さな虫を、指で軽く弾く。",
|
| 35 |
"ギターを弾く銀髪の彼は何人だ?",
|
| 36 |
"この美しい紅葉の絶景を独り占めすることなど、何人たりとも許されない。",
|
|
@@ -46,7 +48,10 @@ def _segments_to_ruby(segments: list[dict]) -> str:
|
|
| 46 |
"""Convert segment dicts to HTML with ruby annotations."""
|
| 47 |
parts: list[str] = []
|
| 48 |
for seg in segments:
|
| 49 |
-
if
|
|
|
|
|
|
|
|
|
|
| 50 |
parts.append(seg["surface"])
|
| 51 |
else:
|
| 52 |
parts.append(
|
|
@@ -63,12 +68,12 @@ CARD_CSS = """\
|
|
| 63 |
.result-container {
|
| 64 |
display: flex;
|
| 65 |
flex-direction: column;
|
| 66 |
-
gap:
|
| 67 |
}
|
| 68 |
.result-card {
|
| 69 |
border: 1px solid var(--border-color-primary);
|
| 70 |
border-radius: 12px;
|
| 71 |
-
padding:
|
| 72 |
background: var(--background-fill-primary);
|
| 73 |
transition: box-shadow 0.2s;
|
| 74 |
}
|
|
@@ -106,13 +111,13 @@ CARD_CSS = """\
|
|
| 106 |
text-align: right;
|
| 107 |
}
|
| 108 |
.ruby-text {
|
| 109 |
-
font-size:
|
| 110 |
line-height: 2.2;
|
| 111 |
letter-spacing: 0.02em;
|
| 112 |
color: var(--body-text-color);
|
| 113 |
}
|
| 114 |
.ruby-text rt {
|
| 115 |
-
font-size: 0.
|
| 116 |
color: var(--color-accent);
|
| 117 |
}
|
| 118 |
.result-card:not(:first-child) .ruby-text {
|
|
@@ -184,7 +189,11 @@ def build_result_html(data: dict) -> str:
|
|
| 184 |
pyopenjtalk_segments = data.get("pyopenjtalk_segments", [])
|
| 185 |
|
| 186 |
if fugashi_segments or pyopenjtalk_segments:
|
| 187 |
-
baselines =
|
|
|
|
|
|
|
|
|
|
|
|
|
| 188 |
if fugashi_segments:
|
| 189 |
fugashi_ruby = _segments_to_ruby(fugashi_segments)
|
| 190 |
baselines += f"""\
|
|
@@ -217,10 +226,10 @@ def build_result_html(data: dict) -> str:
|
|
| 217 |
|
| 218 |
TITLE_HTML = """\
|
| 219 |
<h1 style="text-align:center;margin-bottom:4px;">
|
| 220 |
-
<ruby>Parayomi<rp>(</rp><rt>ぱらよみ</rt><rp>)</rp></ruby> ふりがな
|
| 221 |
</h1>
|
| 222 |
-
<p style="text-align:center;
|
| 223 |
-
日本語の漢字仮名混じり文を入力すると、文脈
|
| 224 |
</p>
|
| 225 |
"""
|
| 226 |
|
|
@@ -232,10 +241,7 @@ def get_client() -> Client:
|
|
| 232 |
global backend_client
|
| 233 |
if backend_client is None:
|
| 234 |
if not backend_space:
|
| 235 |
-
raise gr.Error(
|
| 236 |
-
"BACKEND_SPACE env var is not set. "
|
| 237 |
-
"Set it to the HF Space name (e.g. 'user/parayomi-backend')."
|
| 238 |
-
)
|
| 239 |
backend_client = Client(backend_space)
|
| 240 |
return backend_client
|
| 241 |
|
|
@@ -252,13 +258,13 @@ def process(text: str) -> str:
|
|
| 252 |
# Define textbox outside Blocks so gr.Examples can reference it above
|
| 253 |
text_input = gr.Textbox(
|
| 254 |
label="日本語テキスト",
|
| 255 |
-
lines=
|
| 256 |
value=DEFAULT_TEXT,
|
| 257 |
max_length=100,
|
| 258 |
)
|
| 259 |
|
| 260 |
with gr.Blocks(
|
| 261 |
-
title="Parayomi ふりがな
|
| 262 |
theme=gr.themes.Ocean(),
|
| 263 |
) as demo:
|
| 264 |
gr.HTML(TITLE_HTML)
|
|
@@ -274,7 +280,7 @@ with gr.Blocks(
|
|
| 274 |
text_input.render()
|
| 275 |
|
| 276 |
with gr.Row():
|
| 277 |
-
submit_btn = gr.Button("
|
| 278 |
|
| 279 |
output_html = gr.HTML(label="結果")
|
| 280 |
|
|
|
|
| 12 |
import os
|
| 13 |
|
| 14 |
import gradio as gr
|
| 15 |
+
import jaconv
|
| 16 |
from gradio_client import Client
|
| 17 |
|
| 18 |
# ---------------------------------------------------------------------------
|
|
|
|
| 25 |
|
| 26 |
EXAMPLES = [
|
| 27 |
DEFAULT_TEXT,
|
| 28 |
+
"人気の絶えない観光地だが、一本裏道に入ると急に人気がなくなる。",
|
| 29 |
+
"大事になる前に話し合うべき",
|
| 30 |
+
"歌が上手な彼女は、交渉事でも常に一枚上手であり、舞台の上手で堂々と振る舞った。",
|
| 31 |
+
"綺麗な絵だけど、これは何で色を塗ったの?",
|
| 32 |
+
"綺麗な絵だけど、これは何で金色を塗ったの?",
|
| 33 |
"金の時計を買うために、一生懸命に金を貯めた。",
|
| 34 |
"あちらの方がお見えになった理由は、皆まで言わずとも分かる。",
|
| 35 |
"天気いいし、皆で表に出て遊ぼ?",
|
|
|
|
| 36 |
"カブトムシの立派な角に止まった小さな虫を、指で軽く弾く。",
|
| 37 |
"ギターを弾く銀髪の彼は何人だ?",
|
| 38 |
"この美しい紅葉の絶景を独り占めすることなど、何人たりとも許されない。",
|
|
|
|
| 48 |
"""Convert segment dicts to HTML with ruby annotations."""
|
| 49 |
parts: list[str] = []
|
| 50 |
for seg in segments:
|
| 51 |
+
if (
|
| 52 |
+
seg["surface"] == seg["yomi"]
|
| 53 |
+
or jaconv.kata2hira(seg["surface"]) == seg["yomi"]
|
| 54 |
+
):
|
| 55 |
parts.append(seg["surface"])
|
| 56 |
else:
|
| 57 |
parts.append(
|
|
|
|
| 68 |
.result-container {
|
| 69 |
display: flex;
|
| 70 |
flex-direction: column;
|
| 71 |
+
gap: 8px;
|
| 72 |
}
|
| 73 |
.result-card {
|
| 74 |
border: 1px solid var(--border-color-primary);
|
| 75 |
border-radius: 12px;
|
| 76 |
+
padding: 12px 16px;
|
| 77 |
background: var(--background-fill-primary);
|
| 78 |
transition: box-shadow 0.2s;
|
| 79 |
}
|
|
|
|
| 111 |
text-align: right;
|
| 112 |
}
|
| 113 |
.ruby-text {
|
| 114 |
+
font-size: 1.5em;
|
| 115 |
line-height: 2.2;
|
| 116 |
letter-spacing: 0.02em;
|
| 117 |
color: var(--body-text-color);
|
| 118 |
}
|
| 119 |
.ruby-text rt {
|
| 120 |
+
font-size: 0.8em;
|
| 121 |
color: var(--color-accent);
|
| 122 |
}
|
| 123 |
.result-card:not(:first-child) .ruby-text {
|
|
|
|
| 189 |
pyopenjtalk_segments = data.get("pyopenjtalk_segments", [])
|
| 190 |
|
| 191 |
if fugashi_segments or pyopenjtalk_segments:
|
| 192 |
+
baselines = (
|
| 193 |
+
'<p style="font-size:0.9em;color:var(--body-text-color-subdued);margin:0 0 8px;">'
|
| 194 |
+
"既存の2つの形態素解析器による読み推定の結果:</p>"
|
| 195 |
+
'<div class="baseline-row">'
|
| 196 |
+
)
|
| 197 |
if fugashi_segments:
|
| 198 |
fugashi_ruby = _segments_to_ruby(fugashi_segments)
|
| 199 |
baselines += f"""\
|
|
|
|
| 226 |
|
| 227 |
TITLE_HTML = """\
|
| 228 |
<h1 style="text-align:center;margin-bottom:4px;">
|
| 229 |
+
<ruby>Parayomi<rp>(</rp><rt>ぱらよみ</rt><rp>)</rp></ruby> — ふりがな予測モデル
|
| 230 |
</h1>
|
| 231 |
+
<p style="text-align:center;margin-top:0;">
|
| 232 |
+
日本語の漢字仮名混じり文を入力すると、Parayomiモデルが文脈を考慮した自然な読みを予測します。
|
| 233 |
</p>
|
| 234 |
"""
|
| 235 |
|
|
|
|
| 241 |
global backend_client
|
| 242 |
if backend_client is None:
|
| 243 |
if not backend_space:
|
| 244 |
+
raise gr.Error("BACKEND_SPACE env var is not set. ")
|
|
|
|
|
|
|
|
|
|
| 245 |
backend_client = Client(backend_space)
|
| 246 |
return backend_client
|
| 247 |
|
|
|
|
| 258 |
# Define textbox outside Blocks so gr.Examples can reference it above
|
| 259 |
text_input = gr.Textbox(
|
| 260 |
label="日本語テキスト",
|
| 261 |
+
lines=1,
|
| 262 |
value=DEFAULT_TEXT,
|
| 263 |
max_length=100,
|
| 264 |
)
|
| 265 |
|
| 266 |
with gr.Blocks(
|
| 267 |
+
title="Parayomi — ふりがな予測モデル",
|
| 268 |
theme=gr.themes.Ocean(),
|
| 269 |
) as demo:
|
| 270 |
gr.HTML(TITLE_HTML)
|
|
|
|
| 280 |
text_input.render()
|
| 281 |
|
| 282 |
with gr.Row():
|
| 283 |
+
submit_btn = gr.Button("予測", variant="primary")
|
| 284 |
|
| 285 |
output_html = gr.HTML(label="結果")
|
| 286 |
|
requirements.txt
CHANGED
|
@@ -1,2 +1,3 @@
|
|
| 1 |
gradio>=5.0.0
|
| 2 |
gradio_client>=1.0.0
|
|
|
|
|
|
| 1 |
gradio>=5.0.0
|
| 2 |
gradio_client>=1.0.0
|
| 3 |
+
jaconv
|