Spaces:
Build error
Build error
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,199 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
|
2 |
+
import gradio as gr
|
3 |
+
from transformers import pipeline
|
4 |
+
from utils import preprocess_text, postprocess_translation, TextProcessor
|
5 |
+
from css import custom_css
|
6 |
+
from cultural_utils import get_cultural_context
|
7 |
+
import logging
|
8 |
+
import sys
|
9 |
+
|
10 |
+
logging.basicConfig(
|
11 |
+
level=logging.INFO,
|
12 |
+
format='%(asctime)s - %(levelname)s - %(message)s',
|
13 |
+
stream=sys.stdout
|
14 |
+
)
|
15 |
+
logger = logging.getLogger(__name__)
|
16 |
+
|
17 |
+
class TranslationService:
|
18 |
+
"""Handles the translation pipeline and text processing."""
|
19 |
+
|
20 |
+
def __init__(self):
|
21 |
+
try:
|
22 |
+
logger.info("Loading translation pipeline...")
|
23 |
+
self.pipeline = pipeline(
|
24 |
+
"translation",
|
25 |
+
model="persiannlp/mt5-small-parsinlu-translation_en_fa",
|
26 |
+
device="cpu"
|
27 |
+
)
|
28 |
+
logger.info("Translation pipeline loaded successfully")
|
29 |
+
except Exception as e:
|
30 |
+
logger.error(f"Failed to load translation pipeline: {str(e)}")
|
31 |
+
raise
|
32 |
+
|
33 |
+
def translate(self, text: str, source_lang: str, target_lang: str) -> dict:
|
34 |
+
"""Translate text between English and Farsi with cultural context."""
|
35 |
+
try:
|
36 |
+
if not text:
|
37 |
+
return {"translation": "", "cultural_context": {"idioms": []}}
|
38 |
+
|
39 |
+
# Validate input
|
40 |
+
is_valid, error_msg = TextProcessor.validate_input(text)
|
41 |
+
if not is_valid:
|
42 |
+
return {"translation": error_msg, "cultural_context": {"idioms": []}}
|
43 |
+
|
44 |
+
# Get cultural context before translation
|
45 |
+
src_lang = "en" if source_lang == "English" else "fa"
|
46 |
+
cultural_context = get_cultural_context(text, src_lang)
|
47 |
+
|
48 |
+
# Preprocess text
|
49 |
+
processed_text = preprocess_text(text)
|
50 |
+
logger.info(f"Translating text from {source_lang} to {target_lang}")
|
51 |
+
|
52 |
+
# Generate translation
|
53 |
+
result = self.pipeline(processed_text)
|
54 |
+
|
55 |
+
# Post-process and return result
|
56 |
+
translated_text = postprocess_translation(result[0]['translation_text'])
|
57 |
+
return {
|
58 |
+
"translation": translated_text,
|
59 |
+
"cultural_context": cultural_context
|
60 |
+
}
|
61 |
+
|
62 |
+
except Exception as e:
|
63 |
+
logger.error(f"Translation error: {str(e)}")
|
64 |
+
return {
|
65 |
+
"translation": "Translation error occurred. Please try again.",
|
66 |
+
"cultural_context": {"idioms": []}
|
67 |
+
}
|
68 |
+
|
69 |
+
class GradioInterface:
|
70 |
+
"""Manages the Gradio web interface components."""
|
71 |
+
|
72 |
+
def __init__(self, translation_service):
|
73 |
+
self.translation_service = translation_service
|
74 |
+
|
75 |
+
def create(self):
|
76 |
+
"""Create and configure the Gradio interface."""
|
77 |
+
with gr.Blocks(css=custom_css) as interface:
|
78 |
+
self._create_header()
|
79 |
+
source_lang, target_lang = self._create_language_controls()
|
80 |
+
self._create_translation_interface(source_lang, target_lang)
|
81 |
+
return interface
|
82 |
+
|
83 |
+
def _create_header(self):
|
84 |
+
"""Create the interface header."""
|
85 |
+
gr.Markdown("""
|
86 |
+
# English-Farsi Translation
|
87 |
+
Culturally-sensitive translation between English and Farsi
|
88 |
+
""")
|
89 |
+
|
90 |
+
def _create_language_controls(self):
|
91 |
+
"""Create language selection controls."""
|
92 |
+
with gr.Row():
|
93 |
+
source_lang = gr.Dropdown(
|
94 |
+
choices=["English", "Farsi"],
|
95 |
+
value="English",
|
96 |
+
label="Source Language"
|
97 |
+
)
|
98 |
+
target_lang = gr.Dropdown(
|
99 |
+
choices=["Farsi", "English"],
|
100 |
+
value="Farsi",
|
101 |
+
label="Target Language"
|
102 |
+
)
|
103 |
+
|
104 |
+
swap_btn = gr.Button("🔄 Swap Languages")
|
105 |
+
swap_btn.click(
|
106 |
+
fn=lambda s, t: (t, s),
|
107 |
+
inputs=[source_lang, target_lang],
|
108 |
+
outputs=[source_lang, target_lang]
|
109 |
+
)
|
110 |
+
return source_lang, target_lang
|
111 |
+
|
112 |
+
def _create_translation_interface(self, source_lang, target_lang):
|
113 |
+
"""Create the translation input/output interface."""
|
114 |
+
with gr.Row():
|
115 |
+
with gr.Column():
|
116 |
+
input_text = gr.Textbox(
|
117 |
+
lines=5,
|
118 |
+
placeholder="Enter text to translate...",
|
119 |
+
label="Input Text"
|
120 |
+
)
|
121 |
+
with gr.Column():
|
122 |
+
output_text = gr.Textbox(
|
123 |
+
lines=5,
|
124 |
+
label="Translation",
|
125 |
+
rtl=True
|
126 |
+
)
|
127 |
+
cultural_notes = gr.Markdown(
|
128 |
+
label="Cultural Context",
|
129 |
+
value="Cultural context annotations will appear here..."
|
130 |
+
)
|
131 |
+
|
132 |
+
def format_cultural_context(result):
|
133 |
+
translation = result["translation"]
|
134 |
+
context = result["cultural_context"]
|
135 |
+
|
136 |
+
if not context["idioms"]:
|
137 |
+
return translation, "No cultural annotations found."
|
138 |
+
|
139 |
+
notes = "### Cultural Context Notes:\n"
|
140 |
+
for idiom, literal, explanation in context["idioms"]:
|
141 |
+
notes += f"- **{idiom}**\n"
|
142 |
+
notes += f" - Literal: {literal}\n"
|
143 |
+
notes += f" - Context: {explanation}\n\n"
|
144 |
+
|
145 |
+
return translation, notes
|
146 |
+
|
147 |
+
translate_btn = gr.Button("Translate", variant="primary")
|
148 |
+
def handle_translation(text, src_lang, tgt_lang):
|
149 |
+
result = self.translation_service.translate(text, src_lang, tgt_lang)
|
150 |
+
translation = result["translation"]
|
151 |
+
context = result["cultural_context"]
|
152 |
+
|
153 |
+
if not context["idioms"]:
|
154 |
+
return translation, "No cultural annotations found."
|
155 |
+
|
156 |
+
notes = "### Cultural Context Notes:\n"
|
157 |
+
for idiom, literal, explanation in context["idioms"]:
|
158 |
+
notes += f"- **{idiom}**\n"
|
159 |
+
notes += f" - Literal: {literal}\n"
|
160 |
+
notes += f" - Context: {explanation}\n\n"
|
161 |
+
|
162 |
+
return translation, notes
|
163 |
+
|
164 |
+
translate_btn.click(
|
165 |
+
fn=handle_translation,
|
166 |
+
inputs=[input_text, source_lang, target_lang],
|
167 |
+
outputs=[output_text, cultural_notes]
|
168 |
+
)
|
169 |
+
|
170 |
+
source_lang.change(
|
171 |
+
fn=lambda lang: gr.update(rtl=lang == "Farsi"),
|
172 |
+
inputs=[source_lang],
|
173 |
+
outputs=[input_text]
|
174 |
+
)
|
175 |
+
|
176 |
+
def main():
|
177 |
+
"""Initialize and launch the translation application."""
|
178 |
+
try:
|
179 |
+
logger.info("Initializing translation service...")
|
180 |
+
translation_service = TranslationService()
|
181 |
+
|
182 |
+
logger.info("Creating Gradio interface...")
|
183 |
+
interface = GradioInterface(translation_service).create()
|
184 |
+
|
185 |
+
interface.launch(
|
186 |
+
server_name="0.0.0.0",
|
187 |
+
server_port=8000,
|
188 |
+
share=True,
|
189 |
+
debug=True,
|
190 |
+
show_error=True
|
191 |
+
)
|
192 |
+
logger.info("Gradio interface started successfully")
|
193 |
+
except Exception as e:
|
194 |
+
logger.error(f"Application startup failed: {str(e)}")
|
195 |
+
raise
|
196 |
+
|
197 |
+
if __name__ == "__main__":
|
198 |
+
main()
|
199 |
+
|