Swiss Postal Addresses loading to ElasticSearch

Load all swiss postal adresses from Swiss Post in an ElasticSearch server instance using Python Panda library and the Elastic Search client.

No warranty for completeness and correctness! Use at your own risk.

Usage

1. Download the latest post_adressdaten from https://service.post.ch/zopa/dlc/app/#!/main
2. Place it in the root directory. In swisspostal-adresses_panda.py, change the filename variable value accordingly and set the correct server adress/credentials in the code.
3. Choose an index name (by default 'swiss-streets')
4. Create the index in elastic search: PUT swiss-streets and the mapping PUT swiss-streets/_mapping {... <see mapping.txt> ...}
5. Execute swisspostal-adresses_panda.py, after a few minutes, you should have the 1.8Mio adresses loaded with raw column names from Swiss Post file.

In ElasticSearch, your can now search adresses, for example using:

 GET swiss-streets/_search{
  "query":{
    "multi_match":{
      "query": "Rue 12 Lac Lausanne",
      "fields": ["STR_BEZ_2K","PLZ","ORT_BEZ_18","HNR"],
      "type": "most_fields"
    }
  },
  "size": 400
}

Limitations

This code illustrates the process of converting the Swiss post adresses file into a Panda dataframe and inserting it into Elastic Search. It has the following limitations

• Not all record types are handled or inserted into the ES index
• The python code assumes you have enough RAM to run it in memory. Otherwise it will fail.
• No id is specified, therefore ES use a generated one. This is a problem to maintain records in the index.

Author

Code by Ludovic Favre