Update normalizer.py

normalizer.py

@@ -4,6 +4,12 @@ import pandas as pd
 from aiogoogletrans import Translator
 from spellchecker import SpellChecker
 from nltk.tokenize import RegexpTokenizer
+from langdetect import detect
+from fake_useragent import UserAgent
+import json
+import requests
+from fastapi import HTTPException
+from main import logger
 
 
 class Normalizer:
@@ -353,7 +359,7 @@ class Normalizer:
             print(f"An error occurred during character repetition removal: {e}")
             return word
         
-    def translate_text(self, text: str) -> str:
+    async def translate_text(self, text: str) -> str:
         """
         Translate the given text to English and return the translated text.
 
@@ -364,19 +370,14 @@ class Normalizer:
         - str: The translated text.
         """
         try:
-            loop = asyncio.get_event_loop()
-            translated_text = (
-                loop.run_until_complete(self._translator.translate(text))
-                .text.lower()
-                .strip()
-            )
+            translated_text = await self._translator.translate(text, dest="en")
+            translated_text = translated_text.text.lower().strip()
         except Exception as e:
             print(f"Text Translation failed: {e}")
-            translated_text = (
-                text.lower().strip()
-            )  # Use original text if translation fails
+            translated_text = text.lower().strip()  # Use original text if translation fails
         return translated_text
 
+
     def check_spelling(self, query: str) -> str:
         """
         Check the spelling of the input query and return the corrected version.
@@ -388,9 +389,8 @@ class Normalizer:
         - str: The corrected query.
         """
         try:
-            # Detect the language of the input query using Google Translate API
-            # input_language = self._translator.detect(query)
-            input_language = "en" if query.encode().isalpha() else "ar"
+            # Detect the language of the input query
+            input_language = detect(query)
 
             # Initialize SpellChecker with detected language, fallback to English if language detection fails
             try:
@@ -423,6 +423,62 @@ class Normalizer:
         except Exception as e:
             print(f"An error occurred during spelling check: {e}")
             return query
+        
+    def query_suggestions(self, query: str) -> str:
+        """
+        Get suggestions for a given query string using Google's Suggest API.
+
+        Parameters:
+        query (str): The query string for which suggestions are to be retrieved.
+
+        Returns:
+        str: The suggested query string or the original query if no suggestions are available.
+
+        Raises:
+        HTTPException: If an HTTP error occurs during the request to Google Suggest API.
+        """
+
+        try:
+            # Detect language of the query
+            lang = detect(query)
+            lang = 'en' if lang != 'ar' else 'ar'
+            
+            # Prepare the query for the URL
+            modified_query = query.replace(" ", "+")
+
+            # Construct the URL with the query and language
+            url = f"http://suggestqueries.google.com/complete/search?output=firefox&gl={lang}&hl=sa&q={modified_query}"
+            
+            # logger.info(f"Google Suggest API URL: {url}")
+            # Generate a random user-agent
+            ua = UserAgent()
+            headers = {"user-agent": ua.chrome}
+            
+            # Make the request to the Google Suggest API
+            response = requests.get(url, headers=headers, verify=True)
+            
+            if response.status_code != 200:
+                raise HTTPException(
+                    status_code=response.status_code,
+                    detail=f"An error occurred during the request to Google Suggest API: {response.text}"
+                )
+            # Parse the response JSON
+            suggestions = json.loads(response.text)
+            # logger.info(f"Google Suggest API Response: {suggestions[1]}")
+
+            # If suggestions are available, return the first one
+            if suggestions[1]:
+                return suggestions[1][0]
+            
+            # If no suggestions are returned, return the original query
+            return query
+        
+        except Exception as e:
+            # If any other exception occurs, raise an HTTPException
+            raise HTTPException(
+                status_code=500,
+                detail=f"An error occurred during the request to Google Suggest API: {e}"
+            )
 
     def clean_text(self, text):
         """