A python wrapper for API Offres d'emploi v2 available on Emploi Store (Pole Emploi).
To use the API, you need to subscribe to the Api Offres d'emploi v2. Here are the steps:
- Create an account on Emploi store
- Go to the dashboard (Tableau de bord) and create an application. You should then have the client ID (Identifiant) and client secret key (Clé secrète).
- Go to catalogue and subscribe to Api Offres d'emploi v2
The package can be installed via pip:
pip install api-offres-emploiTo authentificate, create an instance of Api with your client id and secrets (you might want to access these two variables through environment variables instead of hardcoding them):
from offres_emploi import Api
client = Api(client_id="<your client id>",
client_secret="<your client secret>")The API Offres d'emploi v2 has two methods :
search: return aggregate count (named filtresPossibles) of the job offers along with the detailed list of these job offers (named resultats). A lot of parameters are available : date, keywords, localization, etc. Extensive information on the website page Rechercher par critères.referentiel: return reference source (more information on the website page Référentiels).
About range and pagination: In the method search, only the first 150 job offers are returned by default by the API (since the default range value is 0-149). There are three constraints regarding the range parameter (see the reference for the parameter 'range'):
- the maximal value for the first element of range is 1000
- the maximal value for the second element of range is 1149
- the number of result for a given requests must be inferior or equals to 150
Hence for a given search, it is possible to get up to 1150 job offers, by modifying the value of the parameter range for each request (0-149 then 150-299, 300-449, etc.). However some search might have much more than 1150 job offers. In this case one has to narrow down the search using additional parameters (for instance: by playing with the parameters minCreationDate and maxCreationDate of the request).
Limit rate : The API limit rate is set to 3 requests per second.
Equivalence between the API Offres d'emploi v2 methods and this module methods: The 2 main methods from Api Offres d'emploi v2 have the same name in the python module (search and referentiel) and are methods of the module class Api. One notable difference is that the output of search of this module has an additional Content-Range entry (describe hereafter).
The API Offres d'emploi v2 has default parameters, so we can make a requests without parameters (will return the latest job offers) :
basic_search = client.search()The output of the module's Api.search method is the same as the Api Offres d'emploi v2 (dictionnary with entries filtresPossibles and resultats) with an additional entry, Content-Range, that gives the first and last index used for the range parameter along with the total number of available job offers for this request. For instance :
basic_search['Content-Range']will output :
{'first_index': '0', 'last_index': '149', 'max_results': '300749'}The other fields (filtresPossibles and resultats) are explored further the section 4.
A more sensible approach would be to search for special keyword over a certain period, for instance the job offers with that include the keyword big data since 2020-03-01 at 12h30.
from offres_emploi.utils import dt_to_str_iso
import datetime
start_dt = datetime.datetime(2020, 3, 1, 12, 30)
end_dt = datetime.datetime.today()
params = {
"motsCles": "data science",
'minCreationDate': dt_to_str_iso(start_dt),
'maxCreationDate': dt_to_str_iso(end_dt)
}
search_on_big_data = client.search(params=params)The full list of parameters for the search request are available in the page Rechercher par critères. These parameters are passed as dictionary to the parameter params. Note that the keys of this dictionnary are in camelCase, as in the API Offres d'emploi v2 specification. Also, we feed a datetime object to the helper function dt_to_str_iso to convert it to a string with the appropriate ISO-8601 format required by the API.
Getting reference source (referentiel in french) is more straightforward since it does not need any parameter ; one just need to specify the desired reference source:
referentiel_metiers = client.referentiel('metiers')It will output the following list of dictionnaries:
[{'code': '1',
'libelle': "Métiers de l'environnement et du développement durable"},
{'code': '2',
'libelle': 'Métiers de la défense et de la sécurité publique (hors logistique, santé, tertiaire, restauration)'},
{'code': '3',
'libelle': "Métiers du patrimoine et de la restauration d'oeuvres d'art"},
{'code': '4', 'libelle': "Métiers de l'intelligence économique"},
{'code': '5', 'libelle': 'Métiers de la recherche'},
...
]The full list of referentiel is available in the docstring of the method or in the Référentiel page of the API.
The output of the Api.search is a dictionary with three entries:
- filtresPossibles
- resultats
- Content-Range
The breakdown of the available job offers into different categories (type of contract, experience, qualification, nature of contract) are available in the filtresPossibles field and the detailed view of the job offers in resultats field.
filters = search_on_big_data['filtresPossibles']
results = search_on_big_data['resultats']
content_range = search_on_big_data['Content-Range']The number of job offers available at this point in time for this search is given by:
content_range['max_results']The helper function filters_to_df is used to convert the field filtresPossibles in a suitable format:
from offres_emploi.utils import filters_to_df
filters_df = filters_to_df(filters)It will output:
filtre valeur_possible nb_resultats
0 typeContrat CDD 3
1 typeContrat CDI 138
2 typeContrat LIB 37
3 experience 0 15
4 experience 1 96
5 experience 2 47
6 experience 3 20
7 qualification 0 4
8 qualification 9 76
9 qualification X 98
10 natureContrat E1 141
11 natureContrat NS 37It is then straightforward to plot these figures using the data visualization library seaborn:
import seaborn as sns
g = sns.FacetGrid(filters_df, col="filtre", sharex=False, sharey=False)
g = g.map(data=sns.barplot, row="valeur_possible", col="nb_resultats")
The detailed view resultats has a more friendly structure and can be pass directly to a pandas.DataFrame constructor. For example, to know the salary offered by the enterprises for this search:
import pandas as pd
results_df = pd.DataFrame(results)
salary_by_enterprise = (
results_df[['entreprise', 'salaire']]
.dropna()
.agg(dict(entreprise=lambda x: x.get('nom'),
salaire=lambda x: x.get('commentaire')))
.dropna(subset=["salaire"])
.loc[lambda df: df.salaire.str.contains("\d+")]
.sort_values("salaire")
)It will output:
entreprise salaire
73 CLEEVEN SE 25 - 45 k€ brut annuel
112 ASTEK 30 - 40 k€ brut annuel
113 ADSERVIO 30 - 50 k€ brut annuel
15 REDLEAN 30 - 50 k€ brut annuel
29 EOLE CONSULTING 30 - 50 k€ brut annuel
66 PHOENIX ISI 32 - 35 k€ brut annuel
12 DGA DRH CPP SDCO 32 - 52 k€ brut annuel
109 EXPERTEAM 35 - 45 k€ brut annuel
47 GROUPE ASTEN 35 - 48 k€ brut annuel
16 ENERGISME 35 - 60 k€ brut annuel
90 VISEO 36 - 50 k€ brut annuel
80 SOCIETE ALTRAN, Bat Teck - E. Golf Park 38 - 42 k€ brut annuel
14 AINABL TECHNOLOGIES FRANCE 38 - 50 k€ brut annuel
68 SOCIETE ALTRAN, Bat Teck - E. Golf Park 40 - 45 k€ brut annuel
28 Katchme 40 - 50 k€ brut annuel
124 HUMAINEA 42 - 45 k€ brut annuel
79 SOCIETE ALTRAN, Bat Teck - E. Golf Park 45 - 48 k€ brut annuel
86 DGA DRH CPP FDCO A partir de 34 k€ brut annuel
114 SILICOM REGION OUEST A partir de 35 k€ brut annuel
17 SNEF A partir de 40 k€ brut annuel
96 NODYA GROUP A partir de 40 k€ brut annuel
89 QUADRA INFORMATIQUE A partir de 45 k€ brut annuel
35 Hunteed A partir de 49 k€ brut annuelNow you know where to apply.