lot1-kickoff/airflow/dags/opensearch_indexes.py

mappings = {}

mappings['datasources'] = {
    "properties": {
        "data_source_classification": {
            "type": "keyword"
        },
        "jurisdiction": {
            "type": "keyword"
        },
        "local_identifier": {
            "type": "keyword"
        },
        "name": {
            "fields": {
                "keyword": {
                    "type": "keyword"
                }
            },
            "type": "text"
        },
        "research_product_type": {
            "type": "keyword"
        },
        "thematic": {
            "type": "boolean"
        },
        "version_control": {
            "type": "boolean"
        },

        # THOSE WERE NOT INFERREd AUTOMATICALLY
        "submission_policy_url": {
            "type": "text"
        },
        "preservation_policy_url": {
            "type": "text"
        },
        "persistent_identity_systems": {
            "type": "nested",
            "properties": {
                "product_type": {
                    "type": "keyword"
                },
                "pid_schemes": {
                    "type": "keyword"
                }
            }
        },
        "research_product_license": {
            "type": "nested",
            "properties": {
                "name": {
                    "type": "text"
                },
                "url": {
                    "type": "text"
                }
            }
        },
        "research_product_access_policy": {
            "type": "keyword"
        },
        "research_metadata_license": {
            "type": "nested",
            "properties": {
                "name": {
                    "type": "text"
                },
                "url": {
                    "type": "text"
                }
            }
        },
        "research_metadata_access_policy": {
            "type": "keyword"
        },
    }
}

mappings['venues'] = {
    "properties": {
        "identifiers": {
            "type": "nested",
            "properties": {
                "scheme": {
                    "type": "keyword"
                },
                "value": {
                    "type": "keyword"
                }
            }
        },
        "local_identifier": {
            "type": "keyword"
        },
        "name": {
            "fields": {
                "keyword": {
                    "type": "keyword"
                }
            },
            "type": "text"
        },
        "publisher": {
            "fields": {
                "keyword": {
                    "type": "keyword"
                }
            },
            "type": "text"
        },
        "type": {
            "type": "keyword"
        }
    }
}

mappings['topics'] = {
    "properties": {
        "identifiers": {
            "type": "nested",
            "properties": {
                "scheme": {
                    "type": "keyword"
                },
                "value": {
                    "type": "keyword"
                }
            }
        },
        "local_identifier": {
            "type": "keyword"
        },
        "name": {
            "fields": {
                "keyword": {
                    "type": "keyword"
                }
            },
            "type": "text"
        }
    }
}

mappings['persons'] = {
    "properties": {
        "family_name": {
            "fields": {
                "keyword": {
                    "type": "keyword"
                }
            },
            "type": "text"
        },
        "full_name": {
            "fields": {
                "keyword": {
                    "type": "keyword"
                }
            },
            "type": "text"
        },
        "given_name": {
            "fields": {
                "keyword": {
                    "type": "keyword"
                }
            },
            "type": "text"
        },
        "identifiers": {
            "type": "nested",
            "properties": {
                "scheme": {
                    "type": "keyword"
                },
                "value": {
                    "type": "keyword"
                }
            }
        },
        "local_identifier": {
            "type": "keyword"
        }
    }
}

mappings['organizations'] = {
    "properties": {
        "country": {
            "type": "keyword"
        },
        "identifiers": {
            "type": "nested",
            "properties": {
                "scheme": {
                    "type": "keyword"
                },
                "value": {
                    "type": "keyword"
                }
            }
        },
        "local_identifier": {
            "type": "keyword"
        },
        "name": {
            "fields": {
                "keyword": {
                    "type": "keyword"
                }
            },
            "type": "text"
        },
        "other_names": {
            "fields": {
                "keyword": {
                    "type": "keyword"
                }
            },
            "type": "text"
        },
        "short_name": {
            "fields": {
                "keyword": {
                    "type": "keyword"
                }
            },
            "type": "text"
        },
        "type": {
            "type": "keyword"
        }
    }
}

mappings['grants'] = {
    "properties": {
        "acronym": {  # TODO: could be keyword only??
            "fields": {
                "keyword": {
                    "type": "keyword"
                }
            },
            "type": "text"
        },
        "currency": {
            "type": "keyword"
        },
        "end_date": {
            "type": "date",
            "ignore_malformed": "true"
        },
        "funded_amount": {
            "type": "double"
        },
        "funder": {
            "fields": {
                "keyword": {
                    "type": "keyword"
                }
            },
            "type": "text"
        },
        "funding_stream": {
            "fields": {
                "keyword": {
                    "type": "keyword"
                }
            },
            "type": "text"
        },
        "grantCode": {
            "type": "keyword"
        },
        "identifiers": {
            "type": "nested",
            "properties": {
                "scheme": {
                    "type": "keyword"
                },
                "value": {
                    "type": "keyword"
                }
            }
        },
        "local_identifier": {
            "type": "keyword"
        },
        "start_date": {
            "type": "date",
            "ignore_malformed": "true"
        },
        "summary": {
            "fields": {
                "keyword": {
                    "type": "keyword"
                }
            },
            "type": "text"
        },
        "title": {
            "fields": {
                "keyword": {
                    "type": "keyword"
                }
            },
            "type": "text"
        },
        "website": {
            "fields": {
                "keyword": {
                    "type": "keyword"
                }
            },
            "type": "text"
        }
    }
}

mappings['products'] = {
    "properties": {
        "abstracts": {
            "type": "object",
            "properties": {
                # TODO: other languages via index templates
                "none": {
                    "type": "text"
                }
            }
        },
        "contributions": {
            "type": "nested",
            "properties": {
                "person": {
                    "type": "nested",
                    "properties": {
                        "full_name": {
                            "fields": {
                                "keyword": {
                                    "type": "keyword"
                                }
                            },
                            "type": "text"
                        },
                        "local_identifier": {
                            "type": "keyword"
                        },
                        "orcid": {
                            "type": "keyword"
                        }
                    }
                },
                "declared_affiliations": {
                    "type": "keyword"  # TODO: ask to Miriam, not mapped automatically
                },
                "rank": {
                    "type": "long"
                },
                "roles": {
                    "type": "keyword"  # TODO: ask to Miriam, not mapped automatically
                }
            }
        },
        "funding": {
            "type": "nested",
            "properties": {
                "code": {
                    "type": "keyword"
                },
                "funder": {
                    "fields": {
                        "keyword": {
                            "type": "keyword"
                        }
                    },
                    "type": "text"
                },
                "local_identifier": {
                    "type": "keyword"
                },
                "title": {
                    "type": "text"
                }
            }
        },
        "identifiers": {
            "type": "nested",
            "properties": {
                "scheme": {
                    "type": "keyword"
                },
                "value": {
                    "type": "keyword"
                }
            }
        },
        "local_identifier": {
            "type": "keyword"
        },
        "manifestations": {
            "type": "object",
            "properties": {
                "access_right": {
                    "type": "keyword"
                },
                "biblio": {
                    "type": "object",
                    "properties": {
                        "end_page": {
                            "type": "text"
                        },
                        "publisher": {
                            "type": "text"
                        },
                        "start_page": {
                            "type": "text"
                        },
                        "volume": {
                            "type": "text"
                        },
                        # TODO unmapped values
                        "issue": {
                            "type": "text"
                        },
                        "edition": {
                            "type": "text"
                        },
                        "number": {
                            "type": "text"
                        },
                        "venue": {
                            "type": "text"
                        },
                        "hosting_data_source": {
                            "type": "text"
                        }
                    }
                },
                "dates": {
                    "type": "nested",
                    "properties": {
                        "type": {
                            "type": "text"
                        },
                        "value": {
                            "type": "date"
                        }
                    }
                },
                #TODO: should be in biblio ???
                "hosting_datasource": {
                    "type": "object",
                    "properties": {
                        "local_identifier": {
                            "fields": {
                                "keyword": {
                                    "type": "keyword",
                                    "ignore_above": 256
                                }
                            },
                            "type": "text"
                        },
                        "name": {
                            "fields": {
                                "keyword": {
                                    "type": "keyword",
                                    "ignore_above": 256
                                }
                            },
                            "type": "text"
                        }
                    }
                },
                "licence": {
                    "type": "text"
                },
                "metadata_curation": {
                    "type": "keyword"
                },
                "peer_review": {
                    "type": "keyword"
                },
                "pid": {
                    "type": "keyword"
                },
                "product_local_type": {
                    "type": "text"
                },
                "product_local_type_schema": {
                    "type": "text"
                },
                "url": {
                    "type": "text"
                },
                "venue": {
                    "type": "nested",
                    "properties": {
                        "local_identifier": {
                            "type": "keyword"
                        },
                        "name": {
                            "type": "text"
                        }
                    }
                }
            }
        },
        "product_type": {
            "type": "keyword"
        },
        "related_products": {
            "type": "nested",
            "properties": {
                "product_list": {
                    "type": "nested",
                    "properties": {
                        "doi": {
                            "type": "keyword"
                        },
                        "local_identifier": {
                            "type": "keyword"
                        },
                        "title": {
                            "type": "text"
                        }
                    }
                },
                "relation_type": {
                    "type": "keyword"
                }
            }
        },
        "relevant_organizations": {
            "type": "nested",
            "properties": {
                "isni": {
                    "type": "keyword"
                },
                "local_identifier": {
                    "type": "keyword"
                },
                "name": {
                    "type": "text"
                },
                "ror": {
                    "type": "keyword"
                },
                "wikidata": {
                    "type": "keyword"
                }
            }
        },
        "titles": {
            "type": "object",
            #TODO: other languages ??
            "properties": {
                "none": {
                    "type": "text"
                }
            }
        },
        "topics": {
            "type": "nested",
            "properties": {
                "provenance": {
                    "type": "nested",
                    "properties": {
                        "trust": {
                            "type": "double"
                        },
                        "type": {
                            "type": "keyword"
                        }
                    }
                },
                "topic": {
                    "type": "nested",
                    "properties": {
                        "local_identifier": {
                            "type": "keyword"
                        },
                        "value": { # TODO name of the topic???
                            "type": "keyword"
                        }
                    }
                }
            }
        }
    }
}