Merge pull request #796 from NatLibFi/update-dependencies-v1.2

Update dependencies for v1.2 release

.github/workflows/cicd.yml

@@ -101,7 +101,7 @@ jobs:
           # download the small English pretrained spaCy model needed by spacy analyzer
           poetry run python -m spacy download en_core_web_sm --upgrade-strategy only-if-needed
         fi
-        poetry run python -m nltk.downloader punkt
+        poetry run python -m nltk.downloader punkt_tab
     - name: Test with pytest
       run: |
         poetry run pytest --cov=./ --cov-report xml

Dockerfile

@@ -28,7 +28,7 @@ RUN echo "Installing dependencies for optional features: $optional_dependencies"
 	&& rm -rf /root/.cache/pypoetry  # No need for cache because of poetry.lock
 
 # Download nltk data
-RUN python -m nltk.downloader punkt -d /usr/share/nltk_data
+RUN python -m nltk.downloader punkt_tab -d /usr/share/nltk_data
 
 # Download spaCy models, if the optional feature was selected
 ARG spacy_models=en_core_web_sm

README.md

@@ -40,7 +40,7 @@ The recommended way is to install Annif from
 
 You will also need NLTK data files:
 
-    python -m nltk.downloader punkt
+    python -m nltk.downloader punkt_tab
 
 Start up the application:
 
@@ -115,7 +115,7 @@ Enter the virtual environment:
 
 You will also need NLTK data files:
 
-    python -m nltk.downloader punkt
+    python -m nltk.downloader punkt_tab
 
 Start up the application:
 

annif/backend/nn_ensemble.py

@@ -12,7 +12,7 @@
 from typing import TYPE_CHECKING, Any
 
 import joblib
-import keras.backend as K
+import keras
 import lmdb
 import numpy as np
 from keras.layers import Add, Dense, Dropout, Flatten, Input, Layer
@@ -99,7 +99,7 @@ class MeanLayer(Layer):
     """Custom Keras layer that calculates mean values along the 2nd axis."""
 
     def call(self, inputs: EagerTensor) -> EagerTensor:
-        return K.mean(inputs, axis=2)
+        return keras.ops.mean(inputs, axis=2)
 
 
 class NNEnsembleBackend(backend.AnnifLearningBackend, ensemble.BaseEnsembleBackend):

pyproject.toml

@@ -30,33 +30,33 @@ classifiers = [
 [tool.poetry.dependencies]
 python = ">=3.9,<3.13"
 
-connexion = { version = "~3.0.5", extras = ["flask", "uvicorn", "swagger-ui"] }
+connexion = { version = "~3.1.0", extras = ["flask", "uvicorn", "swagger-ui"] }
 click = "8.1.*"
 click-log = "0.4.*"
 joblib = "1.4.*"
-nltk = "3.8.*"
+nltk = "~3.9.1"
 gensim = "4.3.*"
-scikit-learn = "1.4.*"
-scipy = "1.12.*"
+scikit-learn = "~1.5.1"
+scipy = "~1.13.1"
 rdflib = "7.0.*"
-requests = "2.31.*"
-gunicorn = "22.0.*"
+requests = "~2.32.3"
+gunicorn = "~23.0.0"
 numpy = "1.26.*"
-optuna = "3.6.*"
+optuna = "~4.0.0"
 python-dateutil = "2.9.*"
 tomli = { version = "2.0.*", python = "<3.11" }
 simplemma = "~1.1.1"
-jsonschema = "4.21.*"
-huggingface-hub = "0.22.*"
+jsonschema = "~4.23.0"
+huggingface-hub = "~0.25.1"
 
 fasttext-wheel = { version = "0.9.2", optional = true }
 voikko = { version = "0.5.*", optional = true }
-tensorflow-cpu = { version = "2.15.*", optional = true, python = "<3.12" }
-lmdb = { version = "1.4.1", optional = true }
+tensorflow-cpu = { version = "~2.17.0", optional = true }
+lmdb = { version = "~1.5.1", optional = true }
 omikuji = { version = "0.5.*", optional = true }
 yake = { version = "0.4.8", optional = true }
 spacy = { version = "3.7.*", optional = true }
-stwfsapy = { version = "0.4.*", optional = true, python = "<3.12" }
+stwfsapy = { version = "~0.5.0", optional = true }
 
 [tool.poetry.dev-dependencies]
 py = "*"

tests/test_backend_nn_ensemble.py

@@ -66,7 +66,7 @@ def test_nn_ensemble_can_set_lr(registry):
         project=project,
     )
     nn_ensemble._create_model(["dummy-en"])
-    assert nn_ensemble._model.optimizer.learning_rate.value() == 0.002
+    assert nn_ensemble._model.optimizer.learning_rate.value == 0.002
 
 
 def test_set_lmdb_map_size(registry, tmpdir):
@@ -109,7 +109,7 @@ def test_nn_ensemble_train_and_learn(registry, tmpdir):
     nn_ensemble.train(document_corpus)
 
     # check adam default learning_rate:
-    assert nn_ensemble._model.optimizer.learning_rate.value() == 0.001
+    assert nn_ensemble._model.optimizer.learning_rate.value == 0.001
 
     datadir = py.path.local(project.datadir)
     assert datadir.join("nn-model.keras").exists()

tests/test_eval.py

@@ -45,17 +45,17 @@ def test_false_negatives():
 
 # DCG@6 example from https://en.wikipedia.org/wiki/Discounted_cumulative_gain
 def test_dcg():
-    y_true = csr_array([3, 2, 3, 0, 1, 2])
-    y_pred = csr_array([6, 5, 4, 3, 2, 1])
+    y_true = csr_array([[3, 2, 3, 0, 1, 2]])
+    y_pred = csr_array([[6, 5, 4, 3, 2, 1]])
     dcg = annif.eval.dcg_score(y_true, y_pred, 6)
     assert dcg > 6.86
     assert dcg < 6.87
 
 
 # iDCG@6 example from https://en.wikipedia.org/wiki/Discounted_cumulative_gain
 def test_dcg_ideal():
-    y_true = csr_array([3, 3, 3, 2, 2, 2, 1, 0])
-    y_pred = csr_array([8, 7, 6, 5, 4, 3, 2, 1])
+    y_true = csr_array([[3, 3, 3, 2, 2, 2, 1, 0]])
+    y_pred = csr_array([[8, 7, 6, 5, 4, 3, 2, 1]])
     dcg = annif.eval.dcg_score(y_true, y_pred, 6)
     assert dcg > 8.74
     assert dcg < 8.75

tests/test_openapi.py

@@ -4,7 +4,10 @@
 import schemathesis
 from hypothesis import settings
 
-schema = schemathesis.from_path("annif/openapi/annif.yaml")
+import annif
+
+cxapp = annif.create_app(config_name="annif.default_config.TestingConfig")
+schema = schemathesis.from_path("annif/openapi/annif.yaml", app=cxapp)
 
 
 @schemathesis.hook("filter_path_parameters")
@@ -18,18 +21,16 @@ def filter_path_parameters(context, path_parameters):
 
 @schema.parametrize()
 @settings(max_examples=10)
-def test_openapi_fuzzy(case, cxapp):
-    response = case.call_asgi(cxapp)
-    case.validate_response(response)
+def test_openapi_fuzzy(case):
+    case.call_and_validate()
 
 
 @pytest.mark.slow
-@schema.parametrize(endpoint="/v1/projects/{project_id}")
+@schema.include(path_regex="/v1/projects/{project_id}").parametrize()
 @settings(max_examples=50)
-def test_openapi_fuzzy_target_dummy_fi(case, cxapp):
+def test_openapi_fuzzy_target_dummy_fi(case):
     case.path_parameters = {"project_id": "dummy-fi"}
-    response = case.call_asgi(cxapp)
-    case.validate_response(response)
+    case.call_and_validate()
 
 
 def test_openapi_cors(app_client):