invenio-vocabularies 9.1.2__py2.py3-none-any.whl

invenio_vocabularies/contrib/names/mappings/os-v1/names/name-v2.0.0.json

@@ -0,0 +1,165 @@
+{
+  "settings": {
+    "analysis": {
+      "char_filter": {
+        "strip_special_chars": {
+          "type": "pattern_replace",
+          "pattern": "[\\p{Punct}\\p{S}]",
+          "replacement": ""
+        }
+      },
+      "analyzer": {
+        "accent_edge_analyzer": {
+          "tokenizer": "standard",
+          "type": "custom",
+          "char_filter": ["strip_special_chars"],
+          "filter": [
+            "lowercasepreserveoriginal",
+            "asciifoldingpreserveoriginal",
+            "edgegrams"
+          ]
+        },
+        "accent_analyzer": {
+          "tokenizer": "standard",
+          "type": "custom",
+          "char_filter": ["strip_special_chars"],
+          "filter": [
+            "lowercasepreserveoriginal",
+            "asciifoldingpreserveoriginal"
+          ]
+        }
+      },
+      "normalizer": {
+        "accent_normalizer": {
+          "type": "custom",
+          "char_filter": ["strip_special_chars"],
+          "filter": [
+            "lowercase",
+            "asciifolding"
+          ]
+        }
+      },
+      "filter": {
+        "lowercasepreserveoriginal": {
+          "type": "lowercase",
+          "preserve_original": true
+        },
+        "asciifoldingpreserveoriginal": {
+          "type": "asciifolding",
+          "preserve_original": true
+        },
+        "edgegrams": {
+          "type": "edge_ngram",
+          "min_gram": 2,
+          "max_gram": 20
+        }
+      }
+    }
+  },
+  "mappings": {
+    "dynamic": "strict",
+    "properties": {
+      "$schema": {
+        "type": "keyword",
+        "index": "false"
+      },
+      "created": {
+        "type": "date"
+      },
+      "updated": {
+        "type": "date"
+      },
+      "indexed_at": {
+        "type": "date"
+      },
+      "uuid": {
+        "type": "keyword"
+      },
+      "version_id": {
+        "type": "integer"
+      },
+      "id": {
+        "type": "keyword"
+      },
+      "tags": {
+        "type": "keyword"
+      },
+      "name_sort": {
+        "type": "keyword"
+      },
+      "name": {
+        "type": "text",
+        "analyzer": "accent_edge_analyzer",
+        "search_analyzer": "accent_analyzer",
+        "copy_to": "name_sort"
+      },
+      "given_name": {
+        "type": "text",
+        "analyzer": "accent_edge_analyzer",
+        "search_analyzer": "accent_analyzer"
+      },
+      "family_name": {
+        "type": "text"
+      },
+      "identifiers": {
+        "properties": {
+          "identifier": {
+            "type": "keyword",
+            "normalizer": "accent_normalizer"
+          },
+          "scheme": {
+            "type": "keyword"
+          }
+        }
+      },
+      "affiliations": {
+        "type": "object",
+        "properties": {
+          "@v": {
+            "type": "keyword"
+          },
+          "id": {
+            "type": "keyword"
+          },
+          "name": {
+            "type": "text",
+            "analyzer": "accent_edge_analyzer",
+            "search_analyzer": "accent_analyzer"
+          },
+          "acronym": {
+            "type": "text",
+            "analyzer": "accent_edge_analyzer",
+            "search_analyzer": "accent_analyzer",
+            "fields": {
+              "keyword": {
+                "type": "keyword",
+                "normalizer": "accent_normalizer"
+              }
+            }
+          }
+        }
+      },
+      "pid": {
+        "type": "object",
+        "properties": {
+          "pk": {
+            "type": "integer"
+          },
+          "pid_type": {
+            "type": "keyword"
+          },
+          "obj_type": {
+            "type": "keyword"
+          },
+          "status": {
+            "type": "keyword"
+          }
+        }
+      },
+      "props": {
+        "type": "object",
+        "dynamic": "true"
+      }
+    }
+  }
+}

invenio_vocabularies/contrib/names/mappings/os-v2/__init__.py

@@ -0,0 +1,9 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2022 CERN.
+#
+# Invenio-Vocabularies is free software; you can redistribute it and/or
+# modify it under the terms of the MIT License; see LICENSE file for more
+# details.
+
+"""Names OpenSearch v2 mappings."""

invenio_vocabularies/contrib/names/mappings/os-v2/names/name-v1.0.0.json

@@ -0,0 +1,101 @@
+{
+  "mappings": {
+    "dynamic": "strict",
+    "properties": {
+      "$schema": {
+        "type": "keyword",
+        "index": "false"
+      },
+      "created": {
+        "type": "date"
+      },
+      "updated": {
+        "type": "date"
+      },
+      "indexed_at": {
+        "type": "date"
+      },
+      "uuid": {
+        "type": "keyword"
+      },
+      "version_id": {
+        "type": "integer"
+      },
+      "id": {
+        "type": "keyword"
+      },
+      "tags": {
+        "type": "keyword"
+      },
+      "name_sort": {
+        "type": "keyword"
+      },
+      "name": {
+        "type": "text",
+        "copy_to": "name_sort",
+        "fields": {
+          "suggest": {
+            "type": "search_as_you_type"
+          }
+        }
+      },
+      "given_name": {
+        "type": "text"
+      },
+      "family_name": {
+        "type": "text"
+      },
+      "identifiers": {
+        "properties": {
+          "identifier": {
+            "type": "keyword",
+            "fields": {
+              "suggest": {
+                "type": "search_as_you_type"
+              }
+            }
+          },
+          "scheme": {
+            "type": "keyword"
+          }
+        }
+      },
+      "affiliations": {
+        "type": "object",
+        "properties": {
+          "@v": {
+            "type": "keyword"
+          },
+          "id": {
+            "type": "keyword"
+          },
+          "name": {
+            "type": "keyword",
+            "fields": {
+              "suggest": {
+                "type": "search_as_you_type"
+              }
+            }
+          }
+        }
+      },
+      "pid": {
+        "type": "object",
+        "properties": {
+          "pk": {
+            "type": "integer"
+          },
+          "pid_type": {
+            "type": "keyword"
+          },
+          "obj_type": {
+            "type": "keyword"
+          },
+          "status": {
+            "type": "keyword"
+          }
+        }
+      }
+    }
+  }
+}

invenio_vocabularies/contrib/names/mappings/os-v2/names/name-v2.0.0.json

@@ -0,0 +1,165 @@
+{
+  "settings": {
+    "analysis": {
+      "char_filter": {
+        "strip_special_chars": {
+          "type": "pattern_replace",
+          "pattern": "[\\p{Punct}\\p{S}]",
+          "replacement": ""
+        }
+      },
+      "analyzer": {
+        "accent_edge_analyzer": {
+          "tokenizer": "standard",
+          "type": "custom",
+          "char_filter": ["strip_special_chars"],
+          "filter": [
+            "lowercasepreserveoriginal",
+            "asciifoldingpreserveoriginal",
+            "edgegrams"
+          ]
+        },
+        "accent_analyzer": {
+          "tokenizer": "standard",
+          "type": "custom",
+          "char_filter": ["strip_special_chars"],
+          "filter": [
+            "lowercasepreserveoriginal",
+            "asciifoldingpreserveoriginal"
+          ]
+        }
+      },
+      "normalizer": {
+        "accent_normalizer": {
+          "type": "custom",
+          "char_filter": ["strip_special_chars"],
+          "filter": [
+            "lowercase",
+            "asciifolding"
+          ]
+        }
+      },
+      "filter": {
+        "lowercasepreserveoriginal": {
+          "type": "lowercase",
+          "preserve_original": true
+        },
+        "asciifoldingpreserveoriginal": {
+          "type": "asciifolding",
+          "preserve_original": true
+        },
+        "edgegrams": {
+          "type": "edge_ngram",
+          "min_gram": 2,
+          "max_gram": 20
+        }
+      }
+    }
+  },
+  "mappings": {
+    "dynamic": "strict",
+    "properties": {
+      "$schema": {
+        "type": "keyword",
+        "index": "false"
+      },
+      "created": {
+        "type": "date"
+      },
+      "updated": {
+        "type": "date"
+      },
+      "indexed_at": {
+        "type": "date"
+      },
+      "uuid": {
+        "type": "keyword"
+      },
+      "version_id": {
+        "type": "integer"
+      },
+      "id": {
+        "type": "keyword"
+      },
+      "tags": {
+        "type": "keyword"
+      },
+      "name_sort": {
+        "type": "keyword"
+      },
+      "name": {
+        "type": "text",
+        "analyzer": "accent_edge_analyzer",
+        "search_analyzer": "accent_analyzer",
+        "copy_to": "name_sort"
+      },
+      "given_name": {
+        "type": "text",
+        "analyzer": "accent_edge_analyzer",
+        "search_analyzer": "accent_analyzer"
+      },
+      "family_name": {
+        "type": "text"
+      },
+      "identifiers": {
+        "properties": {
+          "identifier": {
+            "type": "keyword",
+            "normalizer": "accent_normalizer"
+          },
+          "scheme": {
+            "type": "keyword"
+          }
+        }
+      },
+      "affiliations": {
+        "type": "object",
+        "properties": {
+          "@v": {
+            "type": "keyword"
+          },
+          "id": {
+            "type": "keyword"
+          },
+          "name": {
+            "type": "text",
+            "analyzer": "accent_edge_analyzer",
+            "search_analyzer": "accent_analyzer"
+          },
+          "acronym": {
+            "type": "text",
+            "analyzer": "accent_edge_analyzer",
+            "search_analyzer": "accent_analyzer",
+            "fields": {
+              "keyword": {
+                "type": "keyword",
+                "normalizer": "accent_normalizer"
+              }
+            }
+          }
+        }
+      },
+      "pid": {
+        "type": "object",
+        "properties": {
+          "pk": {
+            "type": "integer"
+          },
+          "pid_type": {
+            "type": "keyword"
+          },
+          "obj_type": {
+            "type": "keyword"
+          },
+          "status": {
+            "type": "keyword"
+          }
+        }
+      },
+      "props": {
+        "type": "object",
+        "dynamic": "true"
+      }
+    }
+  }
+}

invenio_vocabularies/contrib/names/mappings/v7/__init__.py

@@ -0,0 +1,9 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2021 CERN.
+#
+# Invenio-Vocabularies is free software; you can redistribute it and/or
+# modify it under the terms of the MIT License; see LICENSE file for more
+# details.
+
+"""Names Elasticsearch v7 mappings."""

invenio_vocabularies/contrib/names/mappings/v7/names/name-v1.0.0.json

@@ -0,0 +1,101 @@
+{
+  "mappings": {
+    "dynamic": "strict",
+    "properties": {
+      "$schema": {
+        "type": "keyword",
+        "index": "false"
+      },
+      "created": {
+        "type": "date"
+      },
+      "updated": {
+        "type": "date"
+      },
+      "indexed_at": {
+        "type": "date"
+      },
+      "uuid": {
+        "type": "keyword"
+      },
+      "version_id": {
+        "type": "integer"
+      },
+      "id": {
+        "type": "keyword"
+      },
+      "tags": {
+        "type": "keyword"
+      },
+      "name_sort": {
+        "type": "keyword"
+      },
+      "name": {
+        "type": "text",
+        "copy_to": "name_sort",
+        "fields": {
+          "suggest": {
+            "type": "search_as_you_type"
+          }
+        }
+      },
+      "given_name": {
+        "type": "text"
+      },
+      "family_name": {
+        "type": "text"
+      },
+      "identifiers": {
+        "properties": {
+          "identifier": {
+            "type": "keyword",
+            "fields": {
+              "suggest": {
+                "type": "search_as_you_type"
+              }
+            }
+          },
+          "scheme": {
+            "type": "keyword"
+          }
+        }
+      },
+      "affiliations": {
+        "type": "object",
+        "properties": {
+          "@v": {
+            "type": "keyword"
+          },
+          "id": {
+            "type": "keyword"
+          },
+          "name": {
+            "type": "keyword",
+            "fields": {
+              "suggest": {
+                "type": "search_as_you_type"
+              }
+            }
+          }
+        }
+      },
+      "pid": {
+        "type": "object",
+        "properties": {
+          "pk": {
+            "type": "integer"
+          },
+          "pid_type": {
+            "type": "keyword"
+          },
+          "obj_type": {
+            "type": "keyword"
+          },
+          "status": {
+            "type": "keyword"
+          }
+        }
+      }
+    }
+  }
+}

invenio_vocabularies/contrib/names/models.py

@@ -0,0 +1,13 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2021 CERN.
+#
+# Invenio-Vocabularies is free software; you can redistribute it and/or
+# modify it under the terms of the MIT License; see LICENSE file for more
+# details.
+
+"""Vocabulary names."""
+
+from .names import record_type
+
+NamesMetadata = record_type.model_cls

invenio_vocabularies/contrib/names/names.py

@@ -0,0 +1,80 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2021-2024 CERN.
+#
+# Invenio-Vocabularies is free software; you can redistribute it and/or
+# modify it under the terms of the MIT License; see LICENSE file for more
+# details.
+
+"""Vocabulary names."""
+
+from flask_resources import JSONSerializer, ResponseHandler
+from invenio_db import db
+from invenio_records.dumpers import SearchDumper
+from invenio_records.dumpers.indexedat import IndexedAtDumperExt
+from invenio_records.dumpers.relations import RelationDumperExt
+from invenio_records.systemfields import ModelField, RelationsField
+from invenio_records_resources.factories.factory import RecordTypeFactory
+from invenio_records_resources.records.systemfields import (
+    ModelPIDField,
+    PIDListRelation,
+)
+from invenio_records_resources.resources.records.headers import etag_headers
+
+from invenio_vocabularies.contrib.names.permissions import NamesPermissionPolicy
+
+from ..affiliations.api import Affiliation
+from .config import NamesSearchOptions, service_components
+from .schema import NameSchema
+
+name_relations = RelationsField(
+    affiliations=PIDListRelation(
+        "affiliations",
+        keys=["name", "acronym"],
+        pid_field=Affiliation.pid,
+        cache_key="affiliations",
+    )
+)
+
+record_type = RecordTypeFactory(
+    "Name",
+    # Data layer
+    pid_field_cls=ModelPIDField,
+    pid_field_kwargs={
+        "model_field_name": "pid",
+    },
+    model_cls_attrs={
+        # cannot set to nullable=False because it would fail at
+        # service level when create({}), see records-resources.
+        "pid": db.Column(db.String(255), unique=True),
+        "internal_id": db.Column(db.String(255), nullable=True, index=True),
+    },
+    schema_version="1.0.0",
+    schema_path="local://names/name-v1.0.0.json",
+    index_name="names-name-v2.0.0",
+    record_cls_attrs={"internal_id": ModelField("internal_id", dump=False)},
+    record_relations=name_relations,
+    record_dumper=SearchDumper(
+        model_fields={"pid": ("id", str)},
+        extensions=[
+            RelationDumperExt("relations"),
+            IndexedAtDumperExt(),
+        ],
+    ),
+    # Service layer
+    service_id="names",
+    service_schema=NameSchema,
+    search_options=NamesSearchOptions,
+    service_components=service_components,
+    permission_policy_cls=NamesPermissionPolicy,
+    # Resource layer
+    endpoint_route="/names",
+    resource_cls_attrs={
+        "response_handlers": {
+            "application/json": ResponseHandler(JSONSerializer(), headers=etag_headers),
+            "application/vnd.inveniordm.v1+json": ResponseHandler(
+                JSONSerializer(), headers=etag_headers
+            ),
+        }
+    },
+)

invenio_vocabularies/contrib/names/permissions.py

@@ -0,0 +1,30 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2020-2024 CERN.
+#
+# Invenio-Vocabularies is free software; you can redistribute it and/or
+# modify it under the terms of the MIT License; see LICENSE file for more
+# details.
+
+"""Vocabulary permissions."""
+
+from invenio_records_permissions.generators import AuthenticatedUser, SystemProcess
+
+from invenio_vocabularies.services.generators import IfTags
+from invenio_vocabularies.services.permissions import PermissionPolicy
+
+
+class NamesPermissionPolicy(PermissionPolicy):
+    """Names permission policy.
+
+    Names endpoints are protected, only authenticated users can access them.
+    """
+
+    can_search = [
+        SystemProcess(),
+        AuthenticatedUser(),
+    ]
+    can_read = [
+        SystemProcess(),
+        IfTags(["unlisted"], then_=[SystemProcess()], else_=[AuthenticatedUser()]),
+    ]