Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,85 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
|
| 3 |
+
import difflib
|
| 4 |
+
|
| 5 |
+
# -----------------------------
|
| 6 |
+
# Load Grammar Correction Model
|
| 7 |
+
# -----------------------------
|
| 8 |
+
MODEL_NAME = "vennify/t5-base-grammar-correction"
|
| 9 |
+
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
|
| 10 |
+
model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_NAME)
|
| 11 |
+
|
| 12 |
+
# -----------------------------
|
| 13 |
+
# Grammar correction function
|
| 14 |
+
# -----------------------------
|
| 15 |
+
def correct_grammar(text, language):
|
| 16 |
+
if not text.strip():
|
| 17 |
+
return "", ""
|
| 18 |
+
|
| 19 |
+
# Prepend instruction
|
| 20 |
+
input_text = "grammar: " + text
|
| 21 |
+
input_ids = tokenizer.encode(input_text, return_tensors="pt")
|
| 22 |
+
|
| 23 |
+
outputs = model.generate(input_ids, max_length=512)
|
| 24 |
+
corrected_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
|
| 25 |
+
|
| 26 |
+
# Highlight differences
|
| 27 |
+
diff = difflib.ndiff(text.split(), corrected_text.split())
|
| 28 |
+
highlighted = []
|
| 29 |
+
for token in diff:
|
| 30 |
+
if token.startswith("+"):
|
| 31 |
+
highlighted.append(f"**{token[2:]}**") # Added / corrected
|
| 32 |
+
elif token.startswith("-"):
|
| 33 |
+
highlighted.append(f"~~{token[2:]}~~") # Removed
|
| 34 |
+
else:
|
| 35 |
+
highlighted.append(token[2:])
|
| 36 |
+
highlighted_text = " ".join(highlighted)
|
| 37 |
+
|
| 38 |
+
return corrected_text, highlighted_text
|
| 39 |
+
|
| 40 |
+
# -----------------------------
|
| 41 |
+
# HF-ready Gradio UI
|
| 42 |
+
# -----------------------------
|
| 43 |
+
with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
| 44 |
+
gr.Markdown("<h1 style='color:#8a2be2'>✨ Grammar Corrector</h1>")
|
| 45 |
+
gr.Markdown("<p style='color:#4b0082'>Correct your English sentences instantly!</p>")
|
| 46 |
+
|
| 47 |
+
with gr.Row():
|
| 48 |
+
with gr.Column(scale=1):
|
| 49 |
+
language = gr.Dropdown(
|
| 50 |
+
choices=["English"],
|
| 51 |
+
value="English",
|
| 52 |
+
label="Language"
|
| 53 |
+
)
|
| 54 |
+
user_input = gr.Textbox(
|
| 55 |
+
label="Enter text (multi-line supported)",
|
| 56 |
+
placeholder="Type your text here...",
|
| 57 |
+
lines=5
|
| 58 |
+
)
|
| 59 |
+
correct_btn = gr.Button("Correct Grammar ✅")
|
| 60 |
+
|
| 61 |
+
with gr.Column(scale=1):
|
| 62 |
+
corrected_output = gr.Textbox(
|
| 63 |
+
label="Corrected Text",
|
| 64 |
+
lines=5
|
| 65 |
+
)
|
| 66 |
+
highlighted_output = gr.Markdown(
|
| 67 |
+
label="Highlighted Changes"
|
| 68 |
+
)
|
| 69 |
+
download_btn = gr.File(label="Download Corrected Text")
|
| 70 |
+
|
| 71 |
+
# Function to generate and save corrected text
|
| 72 |
+
def generate_and_save(text, language):
|
| 73 |
+
corrected, highlighted = correct_grammar(text, language)
|
| 74 |
+
with open("corrected_text.txt", "w", encoding="utf-8") as f:
|
| 75 |
+
f.write(corrected)
|
| 76 |
+
return corrected, highlighted, "corrected_text.txt"
|
| 77 |
+
|
| 78 |
+
correct_btn.click(
|
| 79 |
+
fn=generate_and_save,
|
| 80 |
+
inputs=[user_input, language],
|
| 81 |
+
outputs=[corrected_output, highlighted_output, download_btn]
|
| 82 |
+
)
|
| 83 |
+
|
| 84 |
+
# Launch app (HF Spaces automatically runs this)
|
| 85 |
+
demo.launch()
|