nlp: python-edsnlp: Update to 0.22.0.

* guix-science/packages/nlp.scm (python-edsnlp)
  [source]: Update to 0.22.0.
  [arguments]<#:test-flags>: Use markers to run fast tests only. Ignore flaky
  tests. Ignore tests requiring additional dependencies in order to decrease
  closure size.
  [native-inputs]: Add python-{configobj,joblib,optuna,ruamel.yaml}.

1 files changed, 21 insertions, 61 deletions

diff --git a/guix-science/packages/nlp.scm b/guix-science/packages/nlp.scm
index fe044a9..a11ded7 100644
--- a/guix-science/packages/nlp.scm
+++ b/guix-science/packages/nlp.scm
@@ -12,6 +12,7 @@
   #:use-module (gnu packages python-science)
   #:use-module (gnu packages python-web)
   #:use-module (gnu packages python-xyz)
+  #:use-module (gnu packages serialization)
   #:use-module (gnu packages time)
   #:use-module (guix-science packages machine-learning)
   #:use-module (guix-science packages python-xyz)
@@ -94,7 +95,7 @@ to generate a synthetic dataset.")
 (define-public python-edsnlp
   (package
     (name "python-edsnlp")
-    (version "0.20.0")
+    (version "0.22.0")
     (source
      (origin
        (method git-fetch)
@@ -103,74 +104,27 @@ to generate a synthetic dataset.")
              (commit (string-append "v" version))))
        (file-name (git-file-name name version))
        (sha256
-        (base32 "0bv1iah83rf9mxwza534gjc6ib7b3dqr5ir91s7n58cb9509gw0w"))))
+        (base32 "0ykbf3h9x5jbmfhmaf3b1f9viblsxb2yk8hm9g17i20wm8an6dc4"))))
     (build-system pyproject-build-system)
     (arguments
      (list
       #:test-flags
       #~(list
+         ;; Run fast tests, see upstream CI setup.
+         "-m"
+         "not (docs or ml or processing)"
+         ;; Ignore flaky tests, may timeout.
+         "-k"
+         "not test_read_shuffle_loop"
+         ;; Requires python-datasets.
+         "--ignore=tests/data/test_huggingface_dataset.py"
          ;; Requires python-polars.
          "--ignore=tests/data/test_polars.py"
          "--deselect=tests/test_entrypoints.py::test_entrypoints"
-         ;; Requires python-build.
-         "--ignore=tests/utils/test_package.py"
          ;; Requires python-pyspark.
-         "--deselect=tests/processing/test_processing.py::test_pipelines"
-         "--deselect=tests/processing/test_processing.py::test_spark_missing_types[eds]"
-         "--deselect=tests/processing/test_processing.py::test_arbitrary_callback"
-         "--deselect=tests/processing/test_processing.py::test_pipelines"
-         "--deselect=tests/processing/test_processing.py::test_spark_missing_types[fr]"
-         "--deselect=tests/processing/test_processing.py::test_arbitrary_callback"
-         "--deselect=tests/processing/test_backends.py::test_end_to_end"
-         "--deselect=tests/processing/test_backends.py::test_generator[spark]"
-         "--deselect=tests/utils/test_spark_dtypes.py::test_infer_schema"
-         "--deselect=tests/data/test_pandas.py::test_read_write[eds]"
-         "--deselect=tests/data/test_spark.py::test_read_write[eds]"
-         ;; Requires python-scikit-learn.
-         "--deselect=tests/pipelines/core/test_endlines.py::test_endlines[eds]"
-         "--deselect=tests/pipelines/core/test_endlines.py::test_normalizer_endlines[eds]"
-         "--deselect=tests/data/test_pandas.py::test_read_write[fr]"
-         "--deselect=tests/data/test_spark.py::test_read_write[fr]"
-         "--deselect=tests/pipelines/core/test_endlines.py::test_endlines[fr]"
-         "--deselect=tests/pipelines/core/test_endlines.py::test_normalizer_endlines[fr]"
-         ;; Yields catalogue.RegistryError: Can't find 'sentencizer' in registry edsnlp.
-         "--deselect=tests/test_pipeline.py::test_spacy_component"
-         ;; Requires external data.
-         "--deselect=tests/pipelines/trainable/test_extractive_qa.py::test_ner"
-         "--deselect=tests/pipelines/trainable/test_ner.py::test_ner"
-         "--deselect=tests/pipelines/trainable/test_span_linker.py::test_span_linker"
-         "--deselect=tests/pipelines/trainable/test_span_qualifier.py::test_span_qualifier"
-         "--deselect=tests/pipelines/trainable/test_transformer.py::test_span_getter"
-         "--deselect=tests/test_pipeline.py::test_validate_config"
-         "--deselect=tests/test_pipeline.py::test_load_config"
-         "--deselect=tests/test_pipeline.py::test_config_validation_error"
-         "--deselect=tests/test_pipeline.py::test_huggingface"
-         "--deselect=tests/test_pipeline.py::test_missing_huggingface"
-         "--deselect=tests/training/test_train.py::test_ner_qualif_train_diff_bert"
-         "--deselect=tests/training/test_train.py::test_ner_qualif_train_same_bert"
-         "--deselect=tests/training/test_train.py::test_qualif_train"
-         "--deselect=tests/training/test_train.py::test_dep_parser_train"
-         "--deselect=tests/data/test_stream.py::test_repr"
-         "--deselect=tests/data/test_stream.py::test_shuffle_frozen_ml_pipeline"
-         "--deselect=tests/data/test_stream.py::test_parallel_preprocess_stop"
-         "--deselect=tests/processing/test_backends.py::test_multiprocessing_backend"
-         "--deselect=tests/processing/test_backends.py::test_multiprocessing_gpu_stub_backend[simple-True]"
-         "--deselect=tests/processing/test_backends.py::test_multiprocessing_gpu_stub_backend[multiprocessing-True]"
-         "--deselect=tests/processing/test_backends.py::test_multiprocessing_gpu_stub_backend[multiprocessing-False]"
-         "--deselect=tests/processing/test_backends.py::test_multiprocessing_gpu_stub_backend[spark-True]"
-         "--deselect=tests/processing/test_backends.py::test_multiprocessing_gpu_stub_multi_cpu_deterministic_backend"
-         "--deselect=tests/processing/test_backends.py::test_multiprocessing_gpu_stub_wait[True]"
-         "--deselect=tests/processing/test_backends.py::test_multiprocessing_gpu_stub_wait[False]"
-         "--deselect=tests/processing/test_backends.py::test_multiprocessing_rb_error"
-         "--deselect=tests/processing/test_backends.py::test_multiprocessing_ml_error"
-         "--deselect=tests/test_pipeline.py::test_sequence"
-         "--deselect=tests/test_pipeline.py::test_disk_serialization"
-         "--deselect=tests/test_pipeline.py::test_torch_module"
-         "--deselect=tests/test_pipeline.py::test_cache"
-         "--deselect=tests/test_pipeline.py::test_select_pipes"
-         "--deselect=tests/test_pipeline.py::test_torch_save"
-         "--deselect=tests/test_pipeline.py::test_parameters"
-         "--deselect=tests/test_pipeline.py::test_repr")
+         "--deselect=tests/data/test_pandas.py::test_read_write"
+         ;; Requires build module.
+         "--ignore=tests/utils/test_package.py")
       #:phases
       #~(modify-phases %standard-phases
           (add-after 'unpack 'relax-requirements
@@ -208,7 +162,13 @@ to generate a synthetic dataset.")
                              python-typing-extensions
                              python-umls-downloader
                              python-xxhash))
-    (native-inputs (list python-cython python-pytest python-setuptools))
+    (native-inputs (list python-configobj
+                         python-cython
+                         python-joblib
+                         python-optuna
+                         python-pytest
+                         python-ruamel.yaml
+                         python-setuptools))
     (home-page "https://aphp.github.io/edsnlp/")
     (synopsis "NLP framework tailored for French clinical notes")
     (description