translation fix

app.py

@@ -5,7 +5,7 @@ import sqlite3
 import logging
 from util import process_json_files
 from gematria import calculate_gematria
-from deep_translator import GoogleTranslator
+from deep_translator import GoogleTranslator, exceptions
 
 logging.basicConfig(level=logging.INFO, format='%(message)s')
 
@@ -25,18 +25,22 @@ def initialize_database():
             book INTEGER,
             title TEXT,
             chapter INTEGER,
-            verse INTEGER
+            verse INTEGER,
+            UNIQUE(gematria_sum, words, book, title, chapter, verse)
         )
     ''')
     conn.commit()
     conn.close()
 
 def insert_phrase_to_db(c, gematria_sum, phrase_candidate, book_id, title, chapter_id, verse_id):
-    logging.info(f"Inserting: {gematria_sum}, {phrase_candidate}, {book_id}, {title}, {chapter_id + 1}, {verse_id + 1}")
-    c.execute('''
-        INSERT INTO results (gematria_sum, words, book, title, chapter, verse)
-        VALUES (?, ?, ?, ?, ?, ?)
-    ''', (gematria_sum, phrase_candidate, book_id, title, chapter_id + 1, verse_id + 1))
+    try:
+        logging.info(f"Inserting: {gematria_sum}, {phrase_candidate}, {book_id}, {title}, {chapter_id + 1}, {verse_id + 1}")
+        c.execute('''
+            INSERT INTO results (gematria_sum, words, book, title, chapter, verse)
+            VALUES (?, ?, ?, ?, ?, ?)
+        ''', (gematria_sum, phrase_candidate, book_id, title, chapter_id + 1, verse_id + 1))
+    except sqlite3.IntegrityError:
+        logging.info(f"Entry already exists: {gematria_sum}, {phrase_candidate}, {book_id}, {title}, {chapter_id + 1}, {verse_id + 1}")
 
 def populate_database(tanach_texts, max_phrase_length=1):
     conn = sqlite3.connect('gematria.db')
@@ -70,13 +74,20 @@ def search_gematria_in_db(gematria_sum):
     ''', (gematria_sum,))
     results = c.fetchall()
     conn.close()
+    logging.info(f"Search results: {results}")
     return results
 
 def translate_phrases(phrases):
     translator = GoogleTranslator(source='auto', target='en')
     translated_phrases = []
     for phrase in phrases:
-        translated_phrases.append(translator.translate(phrase))
+        try:
+            logging.info(f"Translating phrase: {phrase}")
+            translated_phrases.append(translator.translate(phrase))
+            logging.info(f"Translated phrase: {translated_phrases[-1]}")
+        except (exceptions.TranslationNotFound, exceptions.NotValidPayload, exceptions.ServerException, exceptions.RequestError) as e:
+            logging.error(f"Error translating phrase '{phrase}': {e}")
+            translated_phrases.append("[Translation Error]")
     return translated_phrases
 
 def db(tanach_texts, max_phrase_length=1):
@@ -103,13 +114,18 @@ def gematria_search_interface(phrase):
         return "No matching phrases found.", "\n".join(debug_output)
 
     phrases = [match[0] for match in matching_phrases]
+    debug_callback(f"Debug: Phrases to be translated: {phrases}")
     translations = translate_phrases(phrases)
+    debug_callback(f"Debug: Translations: {translations}")
 
     result = "Matching phrases:\n"
     for match, translation in zip(matching_phrases, translations):
+        if len(match) < 5:
+            debug_callback(f"Error: Expected tuple of length 5, but got {len(match)}: {match}")
+            continue
         result += f"Book: {match[2]} ({match[3]})\nChapter: {match[4]}, Verse: {match[5]}\nPhrase: {match[0]}\nTranslation: {translation}\n\n"
     
-    return result, "\n".join(debug_callback)
+    return result, "\n".join(debug_output)
 
 def run_test():
     debug_output = []
@@ -131,7 +147,7 @@ def run_test():
     # Load the test JSON contents for 01.json
     test_texts_01 = process_json_files(1, 1)
     db(test_texts_01, max_phrase_length=2)  # Populate the database with 1-word phrases
-    search_phrase_01 = "אתקלך שמעתי"
+    search_phrase_01 = "אלוף דישן"
     expected_gematria_01 = calculate_gematria(search_phrase_01.replace(" ", ""))
 
     matching_phrases_01 = search_gematria_in_db(expected_gematria_01)
@@ -142,8 +158,8 @@ def run_test():
 
 iface = gr.Interface(
     fn=gematria_search_interface,
-    inputs=gr.inputs.Textbox(label="Enter phrase"),
-    outputs=[gr.outputs.Textbox(label="Results"), gr.outputs.Textbox(label="Debug Output")],
+    inputs=gr.Textbox(label="Enter phrase"),
+    outputs=[gr.Textbox(label="Results"), gr.Textbox(label="Debug Output")],
     title="Gematria Search in Tanach",
     description="Search for phrases in Tanach that have the same gematria value as the entered phrase.",
     live=False,  # Disable live update

gematria.db

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:830a2259ac32c26a72c0dcfc88dac57965a299afd28140f50e17e1349f58e73c
-size 1363968
+oid sha256:a74da41e964fe494ca3378faaa3be77929f1b0a680c312cee73106b96fb31055
+size 2891776