refacto: we now require python>=3.10, pydantic>=2 and spacy>=3.8.7

Author: percevalw

.github/workflows/documentation.yml

@@ -13,13 +13,13 @@ env:
 
 jobs:
   Documentation:
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
     steps:
     - uses: actions/checkout@v2
 
     - uses: actions/setup-python@v4
       with:
-        python-version: "3.9"
+        python-version: "3.11"
         cache: 'pip'
 
     - run: echo WEEK=$(date +%V) >>$GITHUB_ENV

.github/workflows/release.yml

@@ -24,7 +24,7 @@ jobs:
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        os: [ubuntu-22.04, windows-latest, macos-latest]
+        os: [ubuntu-latest, windows-latest, macos-latest]
 
     steps:
       - uses: actions/checkout@v4
@@ -42,7 +42,7 @@ jobs:
 
   build_sdist:
     name: Build source distribution
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v2
 
@@ -58,7 +58,7 @@ jobs:
     name: Upload to PyPI
 
     needs: [build_wheels, build_sdist]
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
 
     steps:
     - uses: actions/download-artifact@v4
@@ -76,13 +76,13 @@ jobs:
         # repository_url: https://test.pypi.org/legacy/
 
   Documentation:
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
     steps:
     - uses: actions/checkout@v3
 
     - uses: actions/setup-python@v4
       with:
-        python-version: "3.9"
+        python-version: "3.11"
         cache: 'pip'
 
     - name: Install dependencies

.github/workflows/test-build.yml

@@ -18,7 +18,7 @@ jobs:
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        os: [ubuntu-22.04, windows-latest, macos-latest]
+        os: [ubuntu-latest, windows-latest, macos-latest]
 
     steps:
       - uses: actions/checkout@v2
@@ -33,7 +33,7 @@ jobs:
 
   build_sdist:
     name: Build source distribution
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v2
 

.github/workflows/tests.yml

@@ -17,7 +17,7 @@ jobs:
   linting:
     name: Linting
     if: github.event_name == 'pull_request'
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v3
         with:
@@ -34,11 +34,11 @@ jobs:
 
   pytest:
     name: Pytest
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
     strategy:
       fail-fast: true
       matrix:
-        python-version: ["3.7", "3.8", "3.9", "3.10", "3.11", "3.12"]
+        python-version: ["3.10", "3.11", "3.12", "3.13"]
     steps:
       - uses: actions/checkout@v2
 
@@ -62,28 +62,28 @@ jobs:
 
       - name: Install dependencies
         run: pip install -e . --group dev
-        if: matrix.python-version != '3.9' && matrix.python-version != '3.10' && matrix.python-version != '3.11' && matrix.python-version != '3.12'
+        if: matrix.python-version != '3.10' && matrix.python-version != '3.12' && matrix.python-version != '3.13' && matrix.python-version != '3.14'
 
       - name: Install dependencies
         run: pip install -e . --group dev --group setup
-        if: matrix.python-version == '3.9'
+        if: matrix.python-version == '3.10'
 
       - name: Install dependencies
-        # skip ML tests for 3.10 and 3.11
+        # skip ML tests for 3.12 and 3.13
         run: pip install -e . --group dev-no-ml
-        if: matrix.python-version == '3.10' || matrix.python-version == '3.11' || matrix.python-version == '3.12'
+        if: matrix.python-version == '3.12' || matrix.python-version == '3.13' || matrix.python-version == '3.14'
 
       - name: Test with Pytest on Python ${{ matrix.python-version }}
         env:
           UMLS_API_KEY: ${{ secrets.UMLS_API_KEY }}
         run: coverage run -m pytest --ignore tests/test_docs.py
-        if: matrix.python-version != '3.9'
+        if: matrix.python-version != '3.10'
 
       - name: Test with Pytest on Python ${{ matrix.python-version }}
         env:
           UMLS_API_KEY: ${{ secrets.UMLS_API_KEY }}
         run: coverage run -m pytest
-        if: matrix.python-version == '3.9'
+        if: matrix.python-version == '3.10'
 
       - name: Upload coverage data
         uses: actions/upload-artifact@v4
@@ -114,7 +114,7 @@ jobs:
 
     - uses: actions/setup-python@v5
       with:
-        python-version: 3.9
+        python-version: "3.11"
         cache: 'pip'
 
     - name: Install dependencies
@@ -181,11 +181,11 @@ jobs:
 
   simple-installation:
     name: Simple installation
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
     strategy:
       fail-fast: true
       matrix:
-        python-version: ["3.7", "3.8", "3.9", "3.10", "3.11", "3.12"]
+        python-version: ["3.10", "3.11", "3.12", "3.13"]
     steps:
       - uses: actions/checkout@v2
 

Makefile

@@ -4,7 +4,7 @@
 
 ### Changed
 
-- Dependencies (dev, docs and setup) are now listed in dependency groups and installable via `pip install -e . --group the_group`
+- :explosion: EDS-NLP now requires Python 3.10 or later.
 
 ## v0.19.0 (2025-10-04)
 

changelog.md

@@ -40,7 +40,7 @@ dependencies = [
     "sentencepiece>=0.1.96"
 ]
 
-[project.optional-dependencies]
+[dependency-groups]
 dev = [
     "dvc>=2.37.0; python_version >= '3.8'",
     "pandas>=1.4.0,<2.0.0; python_version >= '3.8'",

docs/tutorials/training-span-classifier.md

@@ -1,6 +1,7 @@
 import contextlib
 import functools
 import importlib
+import importlib.metadata as importlib_metadata
 import os
 import re
 import shutil
@@ -30,20 +31,25 @@
     Union,
 )
 
-import pkg_resources
 import requests
 import spacy
 import srsly
 from confit import Config
 from confit.errors import ConfitValidationError, patch_errors
 from confit.utils.xjson import Reference
+from packaging.requirements import Requirement
+from packaging.version import Version
+from pydantic_core import core_schema
 from spacy.language import BaseDefaults
 from spacy.tokenizer import Tokenizer
 from spacy.tokens import Doc
 from spacy.util import get_lang_class
 from spacy.vocab import Vocab, create_vocab
 from typing_extensions import Literal, Self
 
+import edsnlp
+from edsnlp.utils.collections import flatten
+
 from ..core.registries import PIPE_META, DraftPipe, FactoryMeta, registry
 from ..utils.collections import (
     FrozenDict,
@@ -58,14 +64,6 @@
 if TYPE_CHECKING:
     import torch
 
-try:
-    import importlib.metadata as importlib_metadata
-except ModuleNotFoundError:
-    import importlib_metadata
-
-import edsnlp
-from edsnlp.utils.collections import flatten
-
 EMPTY_LIST = FrozenList()
 FORBIDDEN_AUTO_HF_OWNERS = {
     "models",
@@ -626,11 +624,8 @@ def disabled(self):
         return FrozenList(self._disabled)
 
     @classmethod
-    def __get_validators__(cls):
-        """
-        Pydantic validators generator
-        """
-        yield cls.validate
+    def __get_pydantic_core_schema__(cls, source, handler):
+        return core_schema.no_info_plain_validator_function(cls.validate)
 
     @classmethod
     def validate(cls, v, config=None):
@@ -1336,10 +1331,13 @@ def load_from_huggingface(
     missing_deps = []
     # Check if the dependencies are installed, with the correct version
     for req in reqs:
-        req_without_extra = re.sub(r"\[.*\]", "", req)
+        r = Requirement(re.sub(r"\[.*\]", "", req))
         try:
-            pkg_resources.require(req_without_extra)
-        except (pkg_resources.VersionConflict, pkg_resources.DistributionNotFound):
+            installed = Version(importlib_metadata.version(r.name))
+        except importlib_metadata.PackageNotFoundError:
+            installed = None
+
+        if installed is None or r.specifier and installed not in r.specifier:
             missing_deps.append(req)
 
     if missing_deps:

edsnlp/core/pipeline.py

@@ -1,3 +1,4 @@
+import importlib.metadata as importlib_metadata
 import inspect
 import types
 from contextlib import contextmanager
@@ -17,20 +18,18 @@
 from weakref import WeakKeyDictionary
 
 import catalogue
+import confit
 import spacy
-from confit import Config, Registry, RegistryCollection, set_default_registry
+import spacy.registrations
+from confit import Config, RegistryCollection, set_default_registry
 from confit.errors import ConfitValidationError, patch_errors
 from confit.registry import Draft
 from spacy.pipe_analysis import validate_attrs
+from spacy.pipeline.factories import register_factories
 
 import edsnlp
 from edsnlp.utils.collections import FrozenDict, FrozenList
 
-try:
-    import importlib.metadata as importlib_metadata
-except ImportError:
-    import importlib_metadata
-
 PIPE_META = WeakKeyDictionary()
 
 
@@ -75,6 +74,13 @@ class FactoryMeta:
 _AUTO_INSTANTIATE_COMPLETE_PIPE_DRAFT = True
 
 
+class Registry(confit.Registry):
+    def get(self, name: str) -> Any:
+        if name.startswith("spacy."):
+            spacy.registrations.populate_registry()
+        return super().get(name)
+
+
 class DraftPipe(Draft[T]):
     def __init__(self, func, kwargs):
         super().__init__(func, kwargs)
@@ -277,10 +283,15 @@ def get(self, name: str) -> Any:
         registry_path = list(self.namespace) + [name]
         registry_spacy_path = ["spacy", "factories", name]
         registry_internal_spacy_path = ["spacy", "internal_factories", name]
+        registry_legacy_spacy_path = ["spacy-legacy", "factories", name]
 
         def check_and_return():
             if not catalogue.check_exists(*registry_path):
-                for path in (registry_spacy_path, registry_internal_spacy_path):
+                for path in (
+                    registry_spacy_path,
+                    registry_internal_spacy_path,
+                    registry_legacy_spacy_path,
+                ):
                     if catalogue.check_exists(*path):
                         func = catalogue._get(path)
                         meta = spacy.Language.get_factory_meta(name)
@@ -301,14 +312,14 @@ def check_and_return():
         # Steps 1 & 2
         func = check_and_return()
         if func is None and self.entry_points:
+            register_factories()
             # Update entry points in case packages lookup paths have changed
             catalogue.AVAILABLE_ENTRY_POINTS = importlib_metadata.entry_points()
             # Otherwise, step 3
             self.get_entry_point(name)
             # Then redo steps 1 & 2
             func = check_and_return()
         if func is None:
-            # Otherwise, step 4
             if hasattr(spacy.registry, "_entry_point_factories"):
                 spacy.registry._entry_point_factories.get_entry_point(name)
                 # Then redo steps 1 & 2

edsnlp/core/registries.py

@@ -19,7 +19,6 @@
     Union,
 )
 
-import pydantic
 import spacy
 from confit.registry import ValidatedFunction
 from spacy.tokenizer import Tokenizer
@@ -90,9 +89,7 @@ def validate_kwargs(func, kwargs):
         model = vd.init_model_instance(
             **{k: v for k, v in kwargs.items() if k in spec.args}
         )
-        fields = (
-            model.__fields__ if pydantic.__version__ < "2" else vd.model.model_fields
-        )
+        fields = vd.model.model_fields
         d = {
             k: v
             for k, v in model.__dict__.items()