Add migration from ods to dcat

[maudetes]

Fix https://github.com/etalab/data.gouv.fr/issues/1086 Update HarvestSource as well as harvested Datasets to migrate to the DCAT catalog endpoint.

[maudetes]

A correction has been made due to some redirect from domain registered in source and results in DCAT. Here is the sample code made for this correction:

from datetime import datetime
from udata.app import create_app
from udata.models import Dataset
from udata.harvest.models import HarvestSource

# Mapping of old domains -> new domains
source_mapping = {
    'https://sdem.opendatasoft.com': 'https://www.opendata56.fr',
    'https://paysdelaloire.opendatasoft.com': 'https://data.paysdelaloire.fr',
    'https://enedis.opendatasoft.com': 'https://data.enedis.fr',
    'http://data.haute-garonne.fr': 'https://data.haute-garonne.fr',
    'http://breizh.opendatasoft.com': 'https://data.bretagne.bzh',
    'http://data.ratp.fr': 'https://data.ratp.fr',
    'http://data.laregion.fr': 'https://data.laregion.fr',
    'https://datainfogreffe.fr': 'https://opendata.datainfogreffe.fr',
    'http://opendata.stif.info': 'https://data.iledefrance-mobilites.fr',
}

def rename(url, domain):
    return url.replace(domain, source_mapping[domain])

app = create_app()

with app.app_context():
    for domain in source_mapping:
        for source in HarvestSource.objects(url__contains=domain, validation__state="accepted"):
            print(source.id)
            # Rename source
            source.url = rename(source.url, domain)
            source.save()

            # Delete new datasets duplicate created with new domain
            for dat in Dataset.objects(
                harvest__source_id=str(source.id),
                harvest__remote_id__contains=source_mapping[domain],
                created_at_internal__gte="2024-01-08"
            ):
                # print(f"Suppression de : {dat.harvest.remote_id}")
                dat.deleted=datetime.now()
                dat.save()

            # Rename old datasets and resources with new domain
            for dat in Dataset.objects(
                harvest__source_id=str(source.id),
                harvest__remote_id__contains=domain,
                created_at_internal__lte="2024-01-08"
            ):
                # print(f"Renommage : {dat.harvest.remote_id} -> {rename(dat.harvest.remote_id, domain)}")
                dat.harvest.remote_id = rename(dat.harvest.remote_id, domain)
                for res in dat.resources:
                    res.url = rename(res.url, domain)
                dat.save()