invenio-vocabularies 2.2.0__tar.gz → 2.2.2__tar.gz

{invenio-vocabularies-2.2.0 → invenio-vocabularies-2.2.2}/CHANGES.rst

@@ -1,5 +1,5 @@
 ..
-    Copyright (C) 2020 CERN.
+    Copyright (C) 2020-2023 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
@@ -8,6 +8,16 @@
 Changes
 =======
 
+Version 2.2.2 (2023-10-06)
+
+- alembic rcp: set explicit dependency on pidstore create table
+
+Version 2.2.1 (2023-10-02)
+
+- facets: change caching strategy by caching each vocabulary by id. Replace
+  lru_cache with invenio-cache to ensure that cache expiration uses a TTL that
+  is correctly computed.
+
 Version 2.2.0 (2023-09-19)
 
 - facets: implement in-memory cache

{invenio-vocabularies-2.2.0/invenio_vocabularies.egg-info → invenio-vocabularies-2.2.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: invenio-vocabularies
-Version: 2.2.0
+Version: 2.2.2
 Summary: "Invenio module for managing vocabularies."
 Home-page: https://github.com/inveniosoftware/invenio-vocabularies
 Author: CERN
@@ -41,7 +41,7 @@ Description: ..
         https://invenio-vocabularies.readthedocs.io/
         
         ..
-            Copyright (C) 2020 CERN.
+            Copyright (C) 2020-2023 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
@@ -50,6 +50,16 @@ Description: ..
         Changes
         =======
         
+        Version 2.2.2 (2023-10-06)
+        
+        - alembic rcp: set explicit dependency on pidstore create table
+        
+        Version 2.2.1 (2023-10-02)
+        
+        - facets: change caching strategy by caching each vocabulary by id. Replace
+          lru_cache with invenio-cache to ensure that cache expiration uses a TTL that
+          is correctly computed.
+        
         Version 2.2.0 (2023-09-19)
         
         - facets: implement in-memory cache

{invenio-vocabularies-2.2.0 → invenio-vocabularies-2.2.2}/invenio_vocabularies/__init__.py

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-# Copyright (C) 2020-2022 CERN.
+# Copyright (C) 2020-2023 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
@@ -10,6 +10,6 @@
 
 from .ext import InvenioVocabularies
 
-__version__ = "2.2.0"
+__version__ = "2.2.2"
 
 __all__ = ("__version__", "InvenioVocabularies")

{invenio-vocabularies-2.2.0 → invenio-vocabularies-2.2.2}/invenio_vocabularies/alembic/55a700f897b6_add_names_and_afiliations_pid_column.py

@@ -14,7 +14,7 @@ from alembic import op
 revision = "55a700f897b6"
 down_revision = "e1146238edd3"
 branch_labels = ()
-depends_on = None
+depends_on = "999c62899c20"
 
 
 pids_table = sa.sql.table(

{invenio-vocabularies-2.2.0 → invenio-vocabularies-2.2.2}/invenio_vocabularies/assets/semantic-ui/js/invenio_vocabularies/src/contrib/forms/Funding/CustomAwardForm.js

@@ -72,6 +72,7 @@ function CustomAwardForm({ deserializeFunder, selectedFunding }) {
         selectOnNavigation={false}
         required
         search={(options) => options}
+        isFocused
         onValueChange={({ formikProps }, selectedFundersArray) => {
           if (selectedFundersArray.length === 1) {
             const selectedFunder = selectedFundersArray[0];

{invenio-vocabularies-2.2.0 → invenio-vocabularies-2.2.2}/invenio_vocabularies/contrib/affiliations/facets.py

@@ -14,20 +14,20 @@ from ...services.facets import VocabularyLabels, lazy_get_label
 class AffiliationsLabels(VocabularyLabels):
     """Fetching of vocabulary labels for facets."""
 
-    def __init__(self, vocabulary, cache=True, cache_ttl=3600, service_name=None):
+    def __init__(self, vocabulary, cache=True, cache_ttl=3600, service_id=None):
         """Initialize the labels.
 
         :param vocabulary: the name of the vocabulary type.
         :param cache: use simple process in-memory cache when True.
         :param cache_ttl: cache expiration in seconds.
-        :param service_name: the name of the registered service to be used
+        :param service_id: the id of the registered service to be used
             when fetching values for the vocabulary.
         """
         super().__init__(
             vocabulary,
             cache=cache,
             cache_ttl=cache_ttl,
-            service_name="affiliations",
+            service_id="affiliations",
         )
         self.fields = ("id", "title")  # not configurable
 

{invenio-vocabularies-2.2.0 → invenio-vocabularies-2.2.2}/invenio_vocabularies/contrib/funders/facets.py

@@ -14,20 +14,20 @@ from ...services.facets import VocabularyLabels, lazy_get_label
 class FundersLabels(VocabularyLabels):
     """Fetching of vocabulary labels for facets."""
 
-    def __init__(self, vocabulary, cache=True, cache_ttl=3600, service_name=None):
+    def __init__(self, vocabulary, cache=True, cache_ttl=3600, service_id=None):
         """Initialize the labels.
 
         :param vocabulary: the name of the vocabulary type.
         :param cache: use simple process in-memory cache when True.
         :param cache_ttl: cache expiration in seconds.
-        :param service_name: the name of the registered service to be used
+        :param service_id: the id of the registered service to be used
             when fetching values for the vocabulary.
         """
         super().__init__(
             vocabulary,
             cache=cache,
             cache_ttl=cache_ttl,
-            service_name="funders",
+            service_id="funders",
         )
         self.fields = ("id", "title", "country")  # not configurable
 

invenio-vocabularies-2.2.2/invenio_vocabularies/services/facets.py

@@ -0,0 +1,114 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2021-2023 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 facets."""
+
+from flask_principal import AnonymousIdentity
+from invenio_cache.decorators import cached_with_expiration
+from invenio_i18n.ext import current_i18n
+from invenio_records_resources.proxies import current_service_registry
+from invenio_records_resources.services.errors import FacetNotFoundError
+from marshmallow_utils.fields.babel import gettext_from_dict
+from speaklater import make_lazy_string
+from sqlalchemy.exc import NoResultFound
+
+from ..proxies import current_service
+
+
+def get_service(service_id):
+    """Get the service object by name.
+
+    It is required to access the registry lazily, to avoid "out of
+    application context" errors.
+    """
+    return current_service_registry.get(service_id) if service_id else current_service
+
+
+def get_vocabs(service_id, type, fields, ids):
+    """Fetch vocabulary values by ids, using the service."""
+    service = get_service(service_id)
+    vocabs = service.read_many(
+        AnonymousIdentity(), type=type, ids=list(ids), fields=list(fields)
+    )
+    return list(vocabs.hits)  # the service returns a generator
+
+
+@cached_with_expiration
+def get_cached_vocab(service_id, type, fields, id_):
+    """Cache vocabulary values by type in-memory."""
+    vocabs = get_vocabs(service_id, type, fields, [id_])
+    return vocabs[0] if vocabs else None
+
+
+def lazy_get_label(vocab_item):
+    """Lazy evaluation of a localized vocabulary label."""
+    params = {"locale": current_i18n.locale, "default_locale": "en"}
+
+    return make_lazy_string(gettext_from_dict, vocab_item, **params)
+
+
+class VocabularyLabels:
+    """Fetching of vocabulary labels for facets."""
+
+    def __init__(
+        self, vocabulary, cache=True, cache_ttl=3600, service_id=None, id_field="id"
+    ):
+        """Initialize the labels.
+
+        :param vocabulary: the name of the vocabulary type.
+        :param cache: use simple process in-memory cache when True.
+        :param cache_ttl: cache expiration in seconds.
+        :param service_id: the id of the registered service to be used
+            when fetching values for the vocabulary.
+        :param id_field: the name of the `id` field.
+        """
+        self.vocabulary = vocabulary
+        self.cache = cache
+        self.cache_ttl = cache_ttl
+        self.fields = ("id", "title")  # not configurable
+        self.service_id = service_id
+        self.id_field = id_field
+
+    def _vocab_to_label(self, vocab):
+        """Returns the label string for a vocabulary entry."""
+        return lazy_get_label(vocab["title"])
+
+    def __call__(self, ids):
+        """Return the mapping when evaluated."""
+        if not ids:
+            return {}
+
+        labels = {}
+        try:
+            if self.cache:
+                for id_ in ids:
+                    vocab = get_cached_vocab(
+                        self.service_id,
+                        self.vocabulary,
+                        self.fields,
+                        id_,
+                        cache_ttl=self.cache_ttl,
+                    )
+                    if not vocab:
+                        continue
+                    labels[vocab[self.id_field]] = self._vocab_to_label(vocab)
+            else:
+                vocab_list = get_vocabs(
+                    self.service_id,
+                    self.vocabulary,
+                    self.fields,
+                    ids,
+                )
+                for vocab in vocab_list:
+                    id_ = vocab[self.id_field]
+                    if id_ in ids:
+                        labels[id_] = self._vocab_to_label(vocab)
+        except NoResultFound:
+            raise FacetNotFoundError(self.vocabulary)
+
+        return labels

{invenio-vocabularies-2.2.0 → invenio-vocabularies-2.2.2/invenio_vocabularies.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: invenio-vocabularies
-Version: 2.2.0
+Version: 2.2.2
 Summary: "Invenio module for managing vocabularies."
 Home-page: https://github.com/inveniosoftware/invenio-vocabularies
 Author: CERN
@@ -41,7 +41,7 @@ Description: ..
         https://invenio-vocabularies.readthedocs.io/
         
         ..
-            Copyright (C) 2020 CERN.
+            Copyright (C) 2020-2023 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
@@ -50,6 +50,16 @@ Description: ..
         Changes
         =======
         
+        Version 2.2.2 (2023-10-06)
+        
+        - alembic rcp: set explicit dependency on pidstore create table
+        
+        Version 2.2.1 (2023-10-02)
+        
+        - facets: change caching strategy by caching each vocabulary by id. Replace
+          lru_cache with invenio-cache to ensure that cache expiration uses a TTL that
+          is correctly computed.
+        
         Version 2.2.0 (2023-09-19)
         
         - facets: implement in-memory cache

{invenio-vocabularies-2.2.0 → invenio-vocabularies-2.2.2}/tests/resources/test_tasks_resources.py

@@ -13,7 +13,6 @@ from unittest.mock import patch
 
 import pytest
 
-from invenio_vocabularies.datastreams.datastreams import StreamEntry
 from invenio_vocabularies.datastreams.readers import BaseReader
 from invenio_vocabularies.datastreams.transformers import BaseTransformer
 from invenio_vocabularies.datastreams.writers import BaseWriter

{invenio-vocabularies-2.2.0 → invenio-vocabularies-2.2.2}/tests/services/test_labels.py

@@ -12,7 +12,7 @@
 import pytest
 
 from invenio_vocabularies.records.api import Vocabulary
-from invenio_vocabularies.services.facets import VocabularyLabels, cached_vocabs
+from invenio_vocabularies.services.facets import VocabularyLabels
 
 
 @pytest.fixture(scope="module")
@@ -75,28 +75,3 @@ def test_vocabulary_label_not_found(example_records):
 
     assert labels["eng"] == "English"
     assert not labels.get("glg")  # search wont fail
-
-
-def test_lru_cache():
-    # reset
-    cached_vocabs.cache_clear()
-    # 1st exec, cache miss
-    assert cached_vocabs(None, "languages", ("eng", "glg"), ("id", "title"), ttl_hash=1)
-    cache_info = cached_vocabs.cache_info()
-    assert cache_info.hits == 0
-    assert cache_info.misses == 1
-    assert cache_info.currsize == 1
-
-    # 2nd exec with same params, cache hit
-    assert cached_vocabs(None, "languages", ("eng", "glg"), ("id", "title"), ttl_hash=1)
-    cache_info = cached_vocabs.cache_info()
-    assert cache_info.hits == 1
-    assert cache_info.misses == 1
-    assert cache_info.currsize == 1
-
-    # 3rd exec with different ttl_hash to expire the cache, cache miss
-    assert cached_vocabs(None, "languages", ("eng", "glg"), ("id", "title"), ttl_hash=2)
-    cache_info = cached_vocabs.cache_info()
-    assert cache_info.hits == 1
-    assert cache_info.misses == 2
-    assert cache_info.currsize == 2

invenio-vocabularies-2.2.0/invenio_vocabularies/services/facets.py

@@ -1,135 +0,0 @@
-# -*- coding: utf-8 -*-
-#
-# Copyright (C) 2021-2022 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.
-
-"""Vocabulary facets."""
-
-import time
-from functools import lru_cache
-
-from flask_principal import AnonymousIdentity
-from invenio_i18n.ext import current_i18n
-from invenio_records_resources.proxies import current_service_registry
-from invenio_records_resources.services.errors import FacetNotFoundError
-from marshmallow_utils.fields.babel import gettext_from_dict
-from speaklater import make_lazy_string
-from sqlalchemy.exc import NoResultFound
-
-from ..proxies import current_service
-
-
-def get_service(service_name):
-    """Get the service object by name.
-
-    It is required to access the registry lazily, to avoid out of
-    application context errors.
-    """
-    return (
-        current_service_registry.get(service_name) if service_name else current_service
-    )
-
-
-def get_vocabs(service_name, type, ids, fields):
-    """Fetch vocabulary values by ids, using the service."""
-    service = get_service(service_name)
-    vocabs = service.read_many(AnonymousIdentity(), type=type, ids=ids, fields=fields)
-    return list(vocabs.hits)  # the service returns a generator
-
-
-def get_ttl_hash(seconds=3600):
-    """Return the same value withing `seconds` time period. Default 1 hour."""
-    return round(time.time() / seconds)
-
-
-@lru_cache(maxsize=32)
-def cached_vocabs(service_name, type, ids, fields, ttl_hash=None):
-    """Cache vocabulary values by type with a simple process in-memory cache.
-
-    This cache is meant to greatly optimize performance when getting the title of
-    each facet, during searches, and to avoid accessing DB/Search each time.
-    This cache suffer of a slower warm up at the first cache miss when each process is
-    restarted. A data store cache could be also used as mitigation, with the drawback
-    of the added network latency.
-
-    `@lru_cache` is threadsafe. `maxsize` is set to 32, assuming that there are not more
-    than 32 (multiple of 2 for optimization) different vocabularies types defined
-    in the instance.
-
-    The ttl_hash kwarg is necessary to make the cache expiring, by changing the hash key
-    after the defined time.
-    """
-    return get_vocabs(service_name, type=type, ids=list(ids), fields=list(fields))
-
-
-def lazy_get_label(vocab_item):
-    """Lazy evaluation of a localized vocabulary label."""
-    params = {"locale": current_i18n.locale, "default_locale": "en"}
-
-    return make_lazy_string(gettext_from_dict, vocab_item, **params)
-
-
-class VocabularyLabels:
-    """Fetching of vocabulary labels for facets."""
-
-    def __init__(
-        self, vocabulary, cache=True, cache_ttl=3600, service_name=None, id_field="id"
-    ):
-        """Initialize the labels.
-
-        :param vocabulary: the name of the vocabulary type.
-        :param cache: use simple process in-memory cache when True.
-        :param cache_ttl: cache expiration in seconds.
-        :param service_name: the name of the registered service to be used
-            when fetching values for the vocabulary.
-        :param id_field: the name of the `id` field.
-        """
-        self.vocabulary = vocabulary
-        self.cache = cache
-        self.cache_ttl = cache_ttl
-        self.fields = ("id", "title")  # not configurable
-        self.service_name = service_name
-        self.id_field = id_field
-
-    def _vocab_to_label(self, vocab):
-        """Returns the label string for a vocabulary entry."""
-        return lazy_get_label(vocab["title"])
-
-    def __call__(self, ids):
-        """Return the mapping when evaluated."""
-        try:
-            if self.cache:
-                vocab_list = cached_vocabs(
-                    self.service_name,
-                    self.vocabulary,
-                    tuple(ids),
-                    self.fields,
-                    ttl_hash=get_ttl_hash(seconds=self.cache_ttl),
-                )
-            else:
-                vocab_list = get_vocabs(
-                    self.service_name,
-                    self.vocabulary,
-                    ids,
-                    list(self.fields),
-                )
-        except NoResultFound:
-            raise FacetNotFoundError(self.vocabulary)
-
-        labels = {}
-        ids = set(ids)
-        seen = set()
-        for vocab in vocab_list:
-            # cannot loop over ids because vocab id is inside each item
-            if len(ids) == len(seen):
-                break
-
-            id_ = vocab[self.id_field]
-            if id_ in ids:
-                labels[id_] = self._vocab_to_label(vocab)
-                seen.add(id_)
-
-        return labels