Merge pull request #213 from defog-ai/abdullah/extend-instructions-tr…

…anslation-for-mysql Extended instructions translation functionality for mysql and validated mysql files
defog-ai · Aug 21, 2024 · 236e8b4 · 236e8b4
2 parents a575d45 + 39e250d
commit 236e8b4
Show file tree

Hide file tree

Showing 5 changed files with 176 additions and 41 deletions.
diff --git a/correct_sql_instructions.ipynb b/correct_sql_instructions.ipynb
@@ -478,33 +478,50 @@
    "source": [
     "# Feel free to modify this cell for future dialects\n",
     "\n",
-    "from utils.dialects import instructions_to_sqlite, instructions_to_tsql\n",
+    "from utils.dialects import instructions_to_sqlite, instructions_to_tsql, instructions_to_mysql\n",
     "\n",
     "if \"instructions\" in df.columns:\n",
     "    if dialect == \"sqlite\":\n",
+    "        df['instructions'] = df['instructions'].fillna(\"\")\n",
     "        df[\"instructions\"] = df.progress_apply(\n",
     "            lambda x: instructions_to_sqlite(x[\"instructions\"]), axis=1\n",
     "        )\n",
     "    elif dialect == \"tsql\":\n",
+    "        df['instructions'] = df['instructions'].fillna(\"\")\n",
     "        df[\"instructions\"] = df.progress_apply(\n",
     "            lambda x: instructions_to_tsql(x[\"instructions\"]), axis=1\n",
     "        )\n",
+    "    elif dialect == \"mysql\":\n",
+    "        print(df['instructions'].nunique())\n",
+    "        df['instructions'] = df['instructions'].fillna(\"\")\n",
+    "        df[\"instructions\"] = df.progress_apply(\n",
+    "            lambda x: instructions_to_mysql(x[\"instructions\"]), axis=1\n",
+    "        )\n",
     "    else:\n",
     "        raise ValueError(f\"Dialect not yet supported for instructions translation. Please add an instructions_to_{dialect} function in utils/dialects.py\")\n",
+    "else:\n",
+    "    print(\"No instructions column in the dataframe\")\n",
     "    \n",
     "if \"full_instructions\" in df.columns:\n",
     "    if dialect == \"sqlite\":\n",
+    "        df['full_instructions'] = df['full_instructions'].fillna(\"\")\n",
     "        df[\"full_instructions\"] = df.progress_apply(\n",
     "            lambda x: instructions_to_sqlite(x[\"full_instructions\"]), axis=1\n",
     "        )\n",
     "    elif dialect == \"tsql\":\n",
+    "        df['full_instructions'] = df['full_instructions'].fillna(\"\")\n",
     "        df[\"full_instructions\"] = df.progress_apply(\n",
     "            lambda x: instructions_to_tsql(x[\"full_instructions\"]), axis=1\n",
     "        )\n",
+    "    elif dialect == \"mysql\":\n",
+    "        df['full_instructions'] = df['full_instructions'].fillna(\"\")\n",
+    "        df[\"full_instructions\"] = df.progress_apply(\n",
+    "            lambda x: instructions_to_mysql(x[\"full_instructions\"]), axis=1\n",
+    "        )\n",
     "    else:\n",
     "        raise ValueError(f\"Dialect not yet supported for instructions translation. Please add an instructions_to_{dialect} function in utils/dialects.py\")\n",
     "else:\n",
-    "    print(\"No instructions column in the dataframe\")"
+    "    print(\"No full_instructions column in the dataframe\")"
    ]
   },
   {
@@ -542,6 +559,16 @@
     "    for i in full_instructions:\n",
     "        print(i, \"\\n\")"
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Update the csv file with the new translated instructions\n",
+    "df.to_csv(csv_file, index=False)"
+   ]
   }
  ],
  "metadata": {