chore: Update SteamStoreCleaner to translate non-English categories and genres

Author: Rohit04121998

notebooks/steam_data_cleaning.ipynb

@@ -28,6 +28,7 @@
     "import pandas as pd\n",
     "from bs4 import BeautifulSoup\n",
     "from db import get_db\n",
+    "from deep_translator import GoogleTranslator\n",
     "from settings import Path\n",
     "from sqlalchemy import text\n",
     "from utils import check_na, print_steam_links\n",
@@ -1659,6 +1660,16 @@
     "    for col in ['categories', 'genres']:\n",
     "        df[col] = df[col].apply(lambda x: ';'.join(item['description'] for item in literal_eval(x)))\n",
     "    \n",
+    "    def translate(text):\n",
+    "        try:\n",
+    "            lang = GoogleTranslator(source=\"auto\", target=\"en\").translate(text)\n",
+    "            return lang\n",
+    "        except Exception as e:\n",
+    "            return text\n",
+    "    \n",
+    "    df.loc[df[\"english\"] == 0, \"genres\"] = df.loc[df[\"english\"] == 0, \"genres\"].apply(translate)\n",
+    "    df.loc[df[\"english\"] == 0, \"categories\"] = df.loc[df[\"english\"] == 0, \"categories\"].apply(translate)\n",
+    "    \n",
     "    return df\n",
     "\n",
     "cat_gen_df = process_categories_and_genres(price_df)\n",