Elixir: extract element in a large structured data

I made a get request with HTTPoison, but now I'm trying to parse the JSON file. For that I'm using Poison library. But I'm having some problems to get an element from the JSON object.

Here my code :

def parse_json do
    IO.puts("\nLet's parse JSON file.")

    url = "https://entreprise.data.gouv.fr/api/sirene/v3/etablissements/?page=1&per_page=1&etat_administratif=A&denomination_usuelle=" <> "SAFENERGY"

    start()
    case get(url) do
      {:ok, %{status_code: 200, body: body}} ->
        IO.puts("Success research.")
        decode!(body)
        #|> Map.to_list() NON-FUNCTIONAL
        #|> Enum.find(&match?(["etablissements" | _], &1)) NON-FUNCTIONAL
        #|> Enum.find_value(fn %{"siren" => siren} -> siren end) NON-FUNCTIONAL
        #|> IO.puts(&(&1.siren)) NON-FUNCTIONAL
        |> IO.inspect()

      {:ok, %{status_code: 404}} ->
        IO.puts("None match between establishment's name and Sirene API.")

      {:ok, %{status_code: 500}} ->
        IO.puts("Nonfunctional Sirene API's server (maintenance...).")

      {:ok, %{status_code: 429}} ->
        IO.puts("Exceeding the maximum call volume (7 requests/s maximum).")

      {:error, %{reason: reason}} ->
        IO.puts("Failure research.")
        IO.inspect(reason)

      _ -> IO.puts("Unknown error! Good luck to find it!")
    end
  end

My output:

%{
  "etablissements" => [
    %{
      "statut_diffusion" => "O",
      "libelle_voie" => "MAS DES PERES",
      "distribution_speciale_2" => nil,
      "geo_ligne" => "G",
      "libelle_voie_2" => nil,
      "unite_legale_id" => 33014385,
      "type_voie_2" => nil,
      "geo_type" => "street",
      "code_postal_2" => nil,
      "libelle_pays_etranger" => nil,
      "indice_repetition_2" => nil,
      "libelle_commune" => "MAUGUIO",
      "siret" => "48944519700036",
      "id" => 70725092,
      "distribution_speciale" => nil,
      "indice_repetition" => nil,
      "siren" => "489445197",
      "complement_adresse" => nil,
      "unite_legale" => %{
        "statut_diffusion" => "O",
        "nic_siege" => "00036",
        "prenom_usuel" => nil,
        "sigle" => nil,
        "denomination" => "SAFENERGY",
        "id" => 33014385,
        "pseudonyme" => nil,
        "nom_usage" => nil,
        "siren" => "489445197",
        "date_dernier_traitement" => "2019-11-13T15:06:19",
        "annee_effectifs" => nil,
        "categorie_entreprise" => "PME",
        "tranche_effectifs" => nil,
        "identifiant_association" => nil,
        "sexe" => nil,
        "prenom_2" => nil,
        "caractere_employeur" => "N",
        "nom" => nil,
        "created_at" => "2020-07-02T02:56:19.773+02:00",
        "economie_sociale_solidaire" => "N",
        "prenom_4" => nil,
        "date_fin" => nil,
        "date_debut" => "2019-07-01",
        "prenom_3" => nil,
        "date_creation" => "2006-04-01",
        "annee_categorie_entreprise" => "2017",
        "denomination_usuelle_2" => nil,
        "denomination_usuelle_3" => nil,
        "denomination_usuelle_1" => nil,
        ...
      },
      "date_dernier_traitement" => "2019-11-13T15:06:19",
      "annee_effectifs" => nil,
      "denomination_usuelle" => "SAFENERGY",
      "code_pays_etranger" => nil,
      "complement_adresse_2" => nil,
      "tranche_effectifs" => nil,
      "enseigne_1" => nil,
      "numero_voie_2" => nil,
      "geo_id" => "34154_b163",
      "activite_principale_registre_metiers" => nil,
      "geo_l5" => nil,
      "caractere_employeur" => "N",
      "geo_l4" => "MAS DES PERES",
      "nic" => "00036",
      "code_postal" => "34130",
      "libelle_cedex_2" => nil,
      "created_at" => "2020-07-02T03:12:45.814+02:00",
      "numero_voie" => "9002",
      "longitude" => "3.967449",
      "type_voie" => nil,
      "date_debut" => "2019-07-01",
      "code_cedex_2" => nil,
      "date_creation" => "2019-07-01",
      "code_commune_2" => nil,
      "libelle_pays_etranger_2" => nil,
      "libelle_commune_etranger" => nil,
      "latitude" => "43.603798",
      "code_cedex" => nil,
      "geo_score" => "0.95",
      ...
    }
  ],
  "meta" => %{
    "page" => 1,
    "per_page" => 1,
    "total_pages" => 1,
    "total_results" => 1
  }
}

Like you can see above (four lines in comments), I'm trying to extract "siren" element for instance but it failed... Am I off the mark ?

Use Access in general and Kernel.get_in/2 here in particular

body
|> decode!()
|> get_in(["etablissements", Access.all(), "unite_legale", "siren"])
#⇒ "0"

With Access you might filter the outcome on each step, get all branches, etc. Here we use Access.all/0 to get to all elements of the list.