Removing execution_count key completely from "output_type": "execute_result" produces Invalid Notebook in github

[grzegorz700]

In the case of removing execution_count (+metadata) while preserving outputs produce situation with removing execution_count key field entirely instead of putting the null value. After pushing it to GitHub it produces error:

Invalid Notebook
'execution_count' is a required property
Using nbformat v5.10.4 and nbconvert v7.16.1

So, the little change in the code should fix it.

Example:

  {
   "cell_type": "code",
-  "execution_count": 100,
+  "execution_count": null,
   "id": "4eea4a7f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2471\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[('TEXT1',\n",
       "  Counter({'middle': 1})),\n",
       " ('TEXT2',\n",
       "  Counter({'middle': 1})),\n",
       " ('TEXT3.',\n",
       "  Counter({'middle': 1}))]"
      ]
     },
-    "execution_count": 100,
-    "metadata": {},
-    "output_type": "execute_result"
+    "output_type": "execute_result",
+    "metadata": {}
    }
   ],
   "source": [
    "# Display XYZ\n",
    "print(len(result))\n",
    "result[:3]"
   ]
  },

It's data displayed when we don't print them, but it's the last reference in the cell.

The probable fix is change from :

nbstripout-fast/src/stripoutlib.rs

Line 175 in 2a574f9

obj.remove("execution_count");

to:

obj["execution_count"] = json!(null);

I will post it as a pull request when I have some time. But feel free to do it instead of me, and test it on your own. And quickly release a new fixed version :)