Spaces:
Sleeping
Sleeping
Commit
·
70e576c
1
Parent(s):
b898e50
progress bar
Browse files
app.py
CHANGED
@@ -8,6 +8,7 @@ from util import process_json_files
|
|
8 |
from gematria import calculate_gematria
|
9 |
from deep_translator import GoogleTranslator, exceptions
|
10 |
from urllib.parse import quote_plus
|
|
|
11 |
|
12 |
# Set up logging
|
13 |
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
|
@@ -56,7 +57,8 @@ def populate_database(tanach_texts, max_phrase_length=1):
|
|
56 |
logging.info("Populating database...")
|
57 |
cursor = conn.cursor()
|
58 |
|
59 |
-
for book_id
|
|
|
60 |
# Check if the book is already processed for this max_phrase_length
|
61 |
cursor.execute('''SELECT max_phrase_length FROM processed_books WHERE book = ?''', (book_id,))
|
62 |
result = cursor.fetchone()
|
@@ -71,7 +73,7 @@ def populate_database(tanach_texts, max_phrase_length=1):
|
|
71 |
continue
|
72 |
|
73 |
title = book_data.get('title', 'Unknown')
|
74 |
-
book_names[book_id] = title
|
75 |
|
76 |
chapters = book_data['text']
|
77 |
for chapter_id, chapter in enumerate(chapters):
|
@@ -136,7 +138,7 @@ def get_translation(phrase):
|
|
136 |
def translate_and_store(phrase):
|
137 |
"""Translates a Hebrew phrase to English using Google Translate and handles potential errors."""
|
138 |
global translator
|
139 |
-
max_retries = 3
|
140 |
retries = 0
|
141 |
|
142 |
while retries < max_retries:
|
@@ -145,7 +147,7 @@ def translate_and_store(phrase):
|
|
145 |
logging.debug(f"Translated phrase: {translation}")
|
146 |
return translation
|
147 |
except (exceptions.TranslationNotFound, exceptions.NotValidPayload,
|
148 |
-
exceptions.ServerException, exceptions.RequestError, requests.exceptions.ConnectionError) as e:
|
149 |
retries += 1
|
150 |
logging.warning(f"Error translating phrase '{phrase}': {e}. Retrying... ({retries}/{max_retries})")
|
151 |
|
@@ -163,11 +165,11 @@ def search_gematria_in_db(gematria_sum):
|
|
163 |
logging.debug(f"Found {len(results)} matching phrases for Gematria: {gematria_sum}")
|
164 |
return results
|
165 |
|
166 |
-
def gematria_search_interface(phrase):
|
167 |
"""The main function for the Gradio interface."""
|
168 |
if not phrase.strip():
|
169 |
-
return "Please enter a phrase."
|
170 |
-
|
171 |
global conn, book_names
|
172 |
conn = sqlite3.connect('gematria.db')
|
173 |
cursor = conn.cursor()
|
@@ -177,7 +179,7 @@ def gematria_search_interface(phrase):
|
|
177 |
|
178 |
matching_phrases = search_gematria_in_db(phrase_gematria)
|
179 |
if not matching_phrases:
|
180 |
-
return "No matching phrases found."
|
181 |
|
182 |
# Sort results by book, chapter, and verse
|
183 |
sorted_phrases = sorted(matching_phrases, key=lambda x: (x[1], x[2], x[3]))
|
@@ -238,7 +240,7 @@ def gematria_search_interface(phrase):
|
|
238 |
</style>
|
239 |
"""
|
240 |
|
241 |
-
return style + "\n".join(results)
|
242 |
|
243 |
def flatten_text(text):
|
244 |
"""Helper function to flatten nested lists into a single list."""
|
@@ -252,21 +254,40 @@ def run_app():
|
|
252 |
initialize_translator()
|
253 |
|
254 |
# Pre-populate the database
|
255 |
-
tanach_texts = process_json_files(1, 39)
|
|
|
|
|
256 |
populate_database(tanach_texts, max_phrase_length=12)
|
257 |
-
tanach_texts = process_json_files(27, 27)
|
258 |
-
populate_database(tanach_texts, max_phrase_length=24)
|
259 |
|
260 |
-
|
261 |
-
|
262 |
-
|
263 |
-
|
264 |
-
|
265 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
266 |
live=False,
|
267 |
-
allow_flagging="never"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
268 |
)
|
269 |
-
iface.launch()
|
270 |
|
271 |
if __name__ == "__main__":
|
272 |
run_app()
|
|
|
8 |
from gematria import calculate_gematria
|
9 |
from deep_translator import GoogleTranslator, exceptions
|
10 |
from urllib.parse import quote_plus
|
11 |
+
from tqdm import tqdm
|
12 |
|
13 |
# Set up logging
|
14 |
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
|
|
|
57 |
logging.info("Populating database...")
|
58 |
cursor = conn.cursor()
|
59 |
|
60 |
+
for book_id in tqdm(range(1, len(tanach_texts) + 1), desc="Populating database", leave=False):
|
61 |
+
book_data = tanach_texts[book_id]
|
62 |
# Check if the book is already processed for this max_phrase_length
|
63 |
cursor.execute('''SELECT max_phrase_length FROM processed_books WHERE book = ?''', (book_id,))
|
64 |
result = cursor.fetchone()
|
|
|
73 |
continue
|
74 |
|
75 |
title = book_data.get('title', 'Unknown')
|
76 |
+
book_names[book_id] = title
|
77 |
|
78 |
chapters = book_data['text']
|
79 |
for chapter_id, chapter in enumerate(chapters):
|
|
|
138 |
def translate_and_store(phrase):
|
139 |
"""Translates a Hebrew phrase to English using Google Translate and handles potential errors."""
|
140 |
global translator
|
141 |
+
max_retries = 3
|
142 |
retries = 0
|
143 |
|
144 |
while retries < max_retries:
|
|
|
147 |
logging.debug(f"Translated phrase: {translation}")
|
148 |
return translation
|
149 |
except (exceptions.TranslationNotFound, exceptions.NotValidPayload,
|
150 |
+
exceptions.ServerException, exceptions.RequestError, requests.exceptions.ConnectionError) as e:
|
151 |
retries += 1
|
152 |
logging.warning(f"Error translating phrase '{phrase}': {e}. Retrying... ({retries}/{max_retries})")
|
153 |
|
|
|
165 |
logging.debug(f"Found {len(results)} matching phrases for Gematria: {gematria_sum}")
|
166 |
return results
|
167 |
|
168 |
+
def gematria_search_interface(phrase, progress_bar):
|
169 |
"""The main function for the Gradio interface."""
|
170 |
if not phrase.strip():
|
171 |
+
return "Please enter a phrase.", progress_bar
|
172 |
+
|
173 |
global conn, book_names
|
174 |
conn = sqlite3.connect('gematria.db')
|
175 |
cursor = conn.cursor()
|
|
|
179 |
|
180 |
matching_phrases = search_gematria_in_db(phrase_gematria)
|
181 |
if not matching_phrases:
|
182 |
+
return "No matching phrases found.", progress_bar
|
183 |
|
184 |
# Sort results by book, chapter, and verse
|
185 |
sorted_phrases = sorted(matching_phrases, key=lambda x: (x[1], x[2], x[3]))
|
|
|
240 |
</style>
|
241 |
"""
|
242 |
|
243 |
+
return style + "\n".join(results), progress_bar
|
244 |
|
245 |
def flatten_text(text):
|
246 |
"""Helper function to flatten nested lists into a single list."""
|
|
|
254 |
initialize_translator()
|
255 |
|
256 |
# Pre-populate the database
|
257 |
+
tanach_texts = process_json_files(1, 39)
|
258 |
+
populate_database(tanach_texts, max_phrase_length=1)
|
259 |
+
tanach_texts = process_json_files(27, 27)
|
260 |
populate_database(tanach_texts, max_phrase_length=12)
|
|
|
|
|
261 |
|
262 |
+
progress_bar = gr.Progress(label="Database Population", visible=False)
|
263 |
+
with gr.Blocks() as iface:
|
264 |
+
with gr.Row():
|
265 |
+
phrase_input = gr.Textbox(label="Enter phrase")
|
266 |
+
results_output = gr.HTML(label="Results")
|
267 |
+
iface.launch()
|
268 |
+
|
269 |
+
# Create a hidden interface for populating the database with a progress bar
|
270 |
+
hidden_iface = gr.Interface(
|
271 |
+
fn=populate_database,
|
272 |
+
inputs=[gr.Textbox(label="Tanach texts"), gr.Textbox(label="Max phrase length")],
|
273 |
+
outputs=[progress_bar],
|
274 |
+
title="Populate database",
|
275 |
+
description="Populate the database with phrases from the Tanach.",
|
276 |
live=False,
|
277 |
+
allow_flagging="never",
|
278 |
+
visible=False
|
279 |
+
)
|
280 |
+
|
281 |
+
# Trigger the database population when the main interface is launched
|
282 |
+
iface.launch(inline=False)
|
283 |
+
hidden_iface.launch(inline=False, share=False, server_port=8081) # Hidden interface
|
284 |
+
|
285 |
+
# Run the search function when the "Search" button is clicked
|
286 |
+
phrase_input.change(
|
287 |
+
fn=gematria_search_interface,
|
288 |
+
inputs=[phrase_input, progress_bar],
|
289 |
+
outputs=[results_output, progress_bar]
|
290 |
)
|
|
|
291 |
|
292 |
if __name__ == "__main__":
|
293 |
run_app()
|