input.json

{
  "page": 1,
  "pages": 2270,
  "limit": 10,
  "total": 22693,
  "items": [
    {
      "address": {
        "city": "cityname first dataset",
        "company_name": "companyname first dataset"
      },
      "amount": 998,
      "items": [
        {
          "description": "first part of first dataset",
          "number": "part number of first part of first dataset"
        }
      ],
      "number": "number of first dataset",
      "service_date": {
        "type": "DEFAULT",
        "date": "2015-11-18"
      },
      "vat_option": null
    },
    {
      "address": {
        "city": "cityname second dataset",
        "company_name": "companyname second dataset"
      },
      "amount": 998,
      "items": [
        {
          "description": "first part of second dataset",
          "number": "part number of first part of second dataset"
        },
        {
          "description": "second part of second dataset",
          "number": "part number of second part of second dataset"
        }
      ],
      "number": "number of second dataset",
      "service_date": {
        "type": "DEFAULT",
        "date": "2015-11-18"
      },
      "vat_option": null
    }
  ]
}

output.csv

StackOverflow 46724816.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Helping to solve StackOverflow 46724816 with Pandas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import json"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Load your JSON file into variable"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with open(\"input.json\") as f:\n",
    "    rawjson = json.load(f)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I want to convert your data to a dictonary (or a list of dictionary) to import into a Pandas DataFrame, visualize (and possible massage) it and export it to CSV.\n",
    "\n",
    "In case you only want to have a single element (`[0]`) you could setup your dictionary the following"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'item_address_city': 'cityname first dataset',\n",
       " 'item_address_company_name': 'companyname first dataset',\n",
       " 'items_amount': 998}"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "masterDict = {\n",
    "    \"item_address_city\": rawjson[\"items\"][0][\"address\"][\"city\"],\n",
    "    \"item_address_company_name\": rawjson[\"items\"][0][\"address\"][\"company_name\"],\n",
    "    \"items_amount\": rawjson[\"items\"][0][\"amount\"]\n",
    "}\n",
    "masterDict"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I reckon you don't want a single element but probably more of the `item`s. So let's create a list of dictionaries."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "data = []\n",
    "for element in rawjson[\"items\"]:\n",
    "    data.append({\n",
    "        \"item_address_city\": element[\"address\"][\"city\"],\n",
    "        \"item_address_company_name\": element[\"address\"][\"company_name\"],\n",
    "        \"items_amount\": element[\"amount\"]\n",
    "    })"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This we can easily import into Pandas (and have a look at it)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>item_address_city</th>\n",
       "      <th>item_address_company_name</th>\n",
       "      <th>items_amount</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>cityname first dataset</td>\n",
       "      <td>companyname first dataset</td>\n",
       "      <td>998</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>cityname second dataset</td>\n",
       "      <td>companyname second dataset</td>\n",
       "      <td>998</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         item_address_city   item_address_company_name  items_amount\n",
       "0   cityname first dataset   companyname first dataset           998\n",
       "1  cityname second dataset  companyname second dataset           998"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.DataFrame(data)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Finally, export to CSV, if still needed."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df.to_csv(\"output.csv\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}

Thank you Andi! The loop helps a lot and will collect the data like I need it.
The dictionaries like address->city and address->company are now clear to me.
The problem now is the multiple item arrays like:

      "items": [
        {
          "description": "first part of second dataset",
          "number": "part number of first part of second dataset"
        },
        {
          "description": "second part of second dataset",
          "number": "part number of second part of second dataset"
        }
      ]

If I include that into the loop like:
"items_items": element["items"]["description"]
Of course that doesn’t work, because the whole "items": [ ] set is a string.

This here is a good start when using pandas, showing the flattening of arrays: https://medium.com/towards-data-science/flattening-json-objects-in-python-f5343c794b10
The result shown there after normalization is, that the hobbies are put each to an own column.

OK so far - got that.
I'd like to have only one column "hobbies" and then create new lines:
https://www.dropbox.com/s/nm24xqgvdrslkl0/john.jpg?dl=0
The this is, that the information like "John" or "Los Angeles" need to be copied to the next line then.

Whoooho... think I got it: http://pandas.pydata.org/pandas-docs/stable/io.html#normalization
Now adapting it to my data set... keep you posted 💃

Okay, so far so good - in general, it is working, currently I have a slight problem with a "distinguishing prefix" - i think because I am in conflict with 'numbers' - seems to be used in pandas itself?
json_normalize is a great way to "massage" the json data - great tip using pandas!

https://stackoverflow.com/questions/46731658/json-normalize-delivers-valueerror-need-of-distinguishing-prefix

Now also with working code example... :D