Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -3,9 +3,11 @@ from transformers import pipeline
|
|
| 3 |
import spacy
|
| 4 |
from gradio_client import Client
|
| 5 |
import re
|
|
|
|
| 6 |
# Initialize models
|
| 7 |
nlp = spacy.load("en_core_web_sm")
|
| 8 |
spell_checker = pipeline("text2text-generation", model="oliverguhr/spelling-correction-english-base")
|
|
|
|
| 9 |
def preprocess_capitalization(text: str) -> str:
|
| 10 |
"""Preprocess input text to handle capitalization rules."""
|
| 11 |
words = text.split(" ")
|
|
@@ -22,6 +24,7 @@ def preprocess_capitalization(text: str) -> str:
|
|
| 22 |
processed_words.append(word) # Leave other words unchanged
|
| 23 |
|
| 24 |
return " ".join(processed_words)
|
|
|
|
| 25 |
def preprocess_text(text: str):
|
| 26 |
"""Process text and return corrections with position information."""
|
| 27 |
result = {
|
|
@@ -30,6 +33,7 @@ def preprocess_text(text: str):
|
|
| 30 |
"entities": [],
|
| 31 |
"tags": []
|
| 32 |
}
|
|
|
|
| 33 |
# Apply capitalization preprocessing
|
| 34 |
capitalized_text = preprocess_capitalization(text)
|
| 35 |
if capitalized_text != text:
|
|
@@ -39,6 +43,7 @@ def preprocess_text(text: str):
|
|
| 39 |
"type": "spell"
|
| 40 |
})
|
| 41 |
text = capitalized_text # Update text for further processing
|
|
|
|
| 42 |
# Transformer spell check - only for words that look misspelled
|
| 43 |
spell_checked = spell_checker(text, max_length=512)[0]['generated_text']
|
| 44 |
if spell_checked != text:
|
|
@@ -53,6 +58,7 @@ def preprocess_text(text: str):
|
|
| 53 |
"corrected": corrected,
|
| 54 |
"type": "spell"
|
| 55 |
})
|
|
|
|
| 56 |
# Add fluency/style suggestions (other suggestions)
|
| 57 |
# Only add if the word isn't already in spell suggestions
|
| 58 |
spell_originals = {s["original"] for s in result["spell_suggestions"]}
|
|
@@ -70,11 +76,14 @@ def preprocess_text(text: str):
|
|
| 70 |
"corrected": word + "!",
|
| 71 |
"type": "other"
|
| 72 |
})
|
|
|
|
| 73 |
# Add entities and tags
|
| 74 |
doc = nlp(text)
|
| 75 |
result["entities"] = [{"text": ent.text, "label": ent.label_} for ent in doc.ents]
|
| 76 |
result["tags"] = [token.text for token in doc if token.text.startswith(('#', '@'))]
|
|
|
|
| 77 |
return text, result
|
|
|
|
| 78 |
def translate_text(text: str):
|
| 79 |
"""Just translate the text without preprocessing."""
|
| 80 |
client = Client("Frenchizer/space_21")
|
|
@@ -90,6 +99,7 @@ def translate_text(text: str):
|
|
| 90 |
return translation, result
|
| 91 |
except Exception as e:
|
| 92 |
return f"Error: {str(e)}", {}
|
|
|
|
| 93 |
def preprocess_and_forward(text: str):
|
| 94 |
"""Process text and forward to translation service."""
|
| 95 |
original_text, preprocessing_result = preprocess_text(text)
|
|
@@ -101,139 +111,24 @@ def preprocess_and_forward(text: str):
|
|
| 101 |
return translation, preprocessing_result
|
| 102 |
except Exception as e:
|
| 103 |
return f"Error: {str(e)}", preprocessing_result
|
|
|
|
| 104 |
def translate_only(text: str):
|
| 105 |
"""Endpoint that only does translation without preprocessing or suggestions."""
|
| 106 |
translation, empty_result = translate_text(text)
|
| 107 |
return translation, empty_result
|
| 108 |
|
| 109 |
-
# Function to format and display suggestions
|
| 110 |
-
def format_suggestions(preprocessing_result):
|
| 111 |
-
if not preprocessing_result:
|
| 112 |
-
return "No suggestions available."
|
| 113 |
-
|
| 114 |
-
formatted_output = []
|
| 115 |
-
|
| 116 |
-
# Format spell suggestions
|
| 117 |
-
if preprocessing_result["spell_suggestions"]:
|
| 118 |
-
formatted_output.append("## Spelling Suggestions")
|
| 119 |
-
for suggestion in preprocessing_result["spell_suggestions"]:
|
| 120 |
-
formatted_output.append(f"• '{suggestion['original']}' → '{suggestion['corrected']}'")
|
| 121 |
-
|
| 122 |
-
# Format other suggestions
|
| 123 |
-
if preprocessing_result["other_suggestions"]:
|
| 124 |
-
formatted_output.append("## Style Suggestions")
|
| 125 |
-
for suggestion in preprocessing_result["other_suggestions"]:
|
| 126 |
-
formatted_output.append(f"• '{suggestion['original']}' → '{suggestion['corrected']}'")
|
| 127 |
-
|
| 128 |
-
# Format entities
|
| 129 |
-
if preprocessing_result["entities"]:
|
| 130 |
-
formatted_output.append("## Detected Entities")
|
| 131 |
-
for entity in preprocessing_result["entities"]:
|
| 132 |
-
formatted_output.append(f"• {entity['text']} ({entity['label']})")
|
| 133 |
-
|
| 134 |
-
# Format tags
|
| 135 |
-
if preprocessing_result["tags"]:
|
| 136 |
-
formatted_output.append("## Detected Tags")
|
| 137 |
-
for tag in preprocessing_result["tags"]:
|
| 138 |
-
formatted_output.append(f"• {tag}")
|
| 139 |
-
|
| 140 |
-
if not formatted_output:
|
| 141 |
-
return "No suggestions available."
|
| 142 |
-
|
| 143 |
-
return "\n".join(formatted_output)
|
| 144 |
-
|
| 145 |
# Gradio interface
|
| 146 |
-
with gr.Blocks(
|
| 147 |
-
gr.
|
| 148 |
-
|
| 149 |
-
|
| 150 |
-
|
| 151 |
-
|
| 152 |
-
with gr.Column():
|
| 153 |
-
input_text_full = gr.Textbox(label="Input Text", placeholder="Enter text to process and translate...", lines=5)
|
| 154 |
-
process_translate_button = gr.Button("Process & Translate", variant="primary")
|
| 155 |
-
|
| 156 |
-
with gr.Column():
|
| 157 |
-
output_text_full = gr.Textbox(label="Translated Text", lines=5)
|
| 158 |
-
suggestions_markdown = gr.Markdown(label="Suggestions")
|
| 159 |
-
|
| 160 |
-
# Connect the full processing pipeline
|
| 161 |
-
def process_and_display(text):
|
| 162 |
-
translation, preprocessing_result = preprocess_and_forward(text)
|
| 163 |
-
suggestions = format_suggestions(preprocessing_result)
|
| 164 |
-
return translation, suggestions
|
| 165 |
-
|
| 166 |
-
process_translate_button.click(
|
| 167 |
-
fn=process_and_display,
|
| 168 |
-
inputs=[input_text_full],
|
| 169 |
-
outputs=[output_text_full, suggestions_markdown]
|
| 170 |
-
)
|
| 171 |
-
|
| 172 |
with gr.Tab("Translation Only"):
|
| 173 |
-
|
| 174 |
-
|
| 175 |
-
|
| 176 |
-
|
| 177 |
-
|
| 178 |
-
with gr.Column():
|
| 179 |
-
output_text_translate = gr.Textbox(label="Translated Text", lines=5)
|
| 180 |
-
|
| 181 |
-
# Connect translation-only pipeline
|
| 182 |
-
translate_button.click(
|
| 183 |
-
fn=translate_only,
|
| 184 |
-
inputs=[input_text_translate],
|
| 185 |
-
outputs=[output_text_translate]
|
| 186 |
-
)
|
| 187 |
-
|
| 188 |
-
with gr.Tab("Text Analysis Only"):
|
| 189 |
-
with gr.Row():
|
| 190 |
-
with gr.Column():
|
| 191 |
-
input_text_analysis = gr.Textbox(label="Input Text", placeholder="Enter text to analyze...", lines=5)
|
| 192 |
-
analyze_button = gr.Button("Analyze Text", variant="primary")
|
| 193 |
-
|
| 194 |
-
with gr.Column():
|
| 195 |
-
analysis_markdown = gr.Markdown(label="Analysis Results")
|
| 196 |
-
|
| 197 |
-
# Connect analysis-only pipeline
|
| 198 |
-
def analyze_only(text):
|
| 199 |
-
_, preprocessing_result = preprocess_text(text)
|
| 200 |
-
suggestions = format_suggestions(preprocessing_result)
|
| 201 |
-
return suggestions
|
| 202 |
-
|
| 203 |
-
analyze_button.click(
|
| 204 |
-
fn=analyze_only,
|
| 205 |
-
inputs=[input_text_analysis],
|
| 206 |
-
outputs=[analysis_markdown]
|
| 207 |
-
)
|
| 208 |
-
|
| 209 |
-
with gr.Tab("Help"):
|
| 210 |
-
gr.Markdown("""
|
| 211 |
-
# How to Use This Tool
|
| 212 |
-
|
| 213 |
-
This application provides three main functionalities:
|
| 214 |
-
|
| 215 |
-
## 1. Full Processing
|
| 216 |
-
- Corrects spelling and capitalization errors
|
| 217 |
-
- Identifies named entities and tags
|
| 218 |
-
- Offers style suggestions
|
| 219 |
-
- Translates the processed text to French
|
| 220 |
-
|
| 221 |
-
## 2. Translation Only
|
| 222 |
-
- Directly translates your text to French without any preprocessing
|
| 223 |
-
- Useful when you just need a quick translation
|
| 224 |
-
|
| 225 |
-
## 3. Text Analysis Only
|
| 226 |
-
- Analyzes your text for errors and improvement opportunities
|
| 227 |
-
- Identifies named entities and tags
|
| 228 |
-
- Doesn't perform any translation
|
| 229 |
-
|
| 230 |
-
## Features
|
| 231 |
-
- **Spelling Correction**: Identifies and corrects spelling errors
|
| 232 |
-
- **Capitalization Fixes**: Corrects improper capitalization while preserving acronyms
|
| 233 |
-
- **Style Suggestions**: Offers improvement suggestions for better writing
|
| 234 |
-
- **Entity Recognition**: Identifies people, organizations, locations, etc.
|
| 235 |
-
- **Tag Detection**: Finds hashtags and mentions in your text
|
| 236 |
-
""")
|
| 237 |
-
|
| 238 |
if __name__ == "__main__":
|
| 239 |
demo.launch()
|
|
|
|
| 3 |
import spacy
|
| 4 |
from gradio_client import Client
|
| 5 |
import re
|
| 6 |
+
|
| 7 |
# Initialize models
|
| 8 |
nlp = spacy.load("en_core_web_sm")
|
| 9 |
spell_checker = pipeline("text2text-generation", model="oliverguhr/spelling-correction-english-base")
|
| 10 |
+
|
| 11 |
def preprocess_capitalization(text: str) -> str:
|
| 12 |
"""Preprocess input text to handle capitalization rules."""
|
| 13 |
words = text.split(" ")
|
|
|
|
| 24 |
processed_words.append(word) # Leave other words unchanged
|
| 25 |
|
| 26 |
return " ".join(processed_words)
|
| 27 |
+
|
| 28 |
def preprocess_text(text: str):
|
| 29 |
"""Process text and return corrections with position information."""
|
| 30 |
result = {
|
|
|
|
| 33 |
"entities": [],
|
| 34 |
"tags": []
|
| 35 |
}
|
| 36 |
+
|
| 37 |
# Apply capitalization preprocessing
|
| 38 |
capitalized_text = preprocess_capitalization(text)
|
| 39 |
if capitalized_text != text:
|
|
|
|
| 43 |
"type": "spell"
|
| 44 |
})
|
| 45 |
text = capitalized_text # Update text for further processing
|
| 46 |
+
|
| 47 |
# Transformer spell check - only for words that look misspelled
|
| 48 |
spell_checked = spell_checker(text, max_length=512)[0]['generated_text']
|
| 49 |
if spell_checked != text:
|
|
|
|
| 58 |
"corrected": corrected,
|
| 59 |
"type": "spell"
|
| 60 |
})
|
| 61 |
+
|
| 62 |
# Add fluency/style suggestions (other suggestions)
|
| 63 |
# Only add if the word isn't already in spell suggestions
|
| 64 |
spell_originals = {s["original"] for s in result["spell_suggestions"]}
|
|
|
|
| 76 |
"corrected": word + "!",
|
| 77 |
"type": "other"
|
| 78 |
})
|
| 79 |
+
|
| 80 |
# Add entities and tags
|
| 81 |
doc = nlp(text)
|
| 82 |
result["entities"] = [{"text": ent.text, "label": ent.label_} for ent in doc.ents]
|
| 83 |
result["tags"] = [token.text for token in doc if token.text.startswith(('#', '@'))]
|
| 84 |
+
|
| 85 |
return text, result
|
| 86 |
+
|
| 87 |
def translate_text(text: str):
|
| 88 |
"""Just translate the text without preprocessing."""
|
| 89 |
client = Client("Frenchizer/space_21")
|
|
|
|
| 99 |
return translation, result
|
| 100 |
except Exception as e:
|
| 101 |
return f"Error: {str(e)}", {}
|
| 102 |
+
|
| 103 |
def preprocess_and_forward(text: str):
|
| 104 |
"""Process text and forward to translation service."""
|
| 105 |
original_text, preprocessing_result = preprocess_text(text)
|
|
|
|
| 111 |
return translation, preprocessing_result
|
| 112 |
except Exception as e:
|
| 113 |
return f"Error: {str(e)}", preprocessing_result
|
| 114 |
+
|
| 115 |
def translate_only(text: str):
|
| 116 |
"""Endpoint that only does translation without preprocessing or suggestions."""
|
| 117 |
translation, empty_result = translate_text(text)
|
| 118 |
return translation, empty_result
|
| 119 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 120 |
# Gradio interface
|
| 121 |
+
with gr.Blocks() as demo:
|
| 122 |
+
with gr.Tab("Main"):
|
| 123 |
+
input_text = gr.Textbox(label="Input Text")
|
| 124 |
+
output_text = gr.Textbox(label="Output Text")
|
| 125 |
+
preprocess_button = gr.Button("Process and Translate")
|
| 126 |
+
preprocess_button.click(fn=preprocess_and_forward, inputs=input_text, outputs=output_text)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 127 |
with gr.Tab("Translation Only"):
|
| 128 |
+
translate_input = gr.Textbox(label="Input Text")
|
| 129 |
+
translate_output = gr.Textbox(label="Output Text")
|
| 130 |
+
translate_button = gr.Button("Translate")
|
| 131 |
+
translate_button.click(fn=translate_only, inputs=translate_input, outputs=translate_output)
|
| 132 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 133 |
if __name__ == "__main__":
|
| 134 |
demo.launch()
|