[project]
name = "edsnlp"
description = "Modular, fast NLP framework, compatible with Pytorch and spaCy, offering tailored support for French clinical notes."
authors = [
{ name = "Data Science - DSN APHP", email = "perceval.wajsburt@aphp.fr" }
]
license = { file = "LICENSE" }
readme = "README.md"
requires-python = ">=3.7.1"
dynamic = ['version']
dependencies = [
"loguru",
"pytz",
"pysimstring>=1.2.1",
"regex",
# spacy doesn't provide binaries for python<3.9 from 3.8.2 so we need to cap it ourself
"spacy>=3.2,<3.8.2; python_version<'3.9'",
"spacy>=3.8.5,<4.0.0; python_version>='3.9'",
# thinc doesn't provide binaries for python<3.9 from 8.2.5 so we need to cap it ourself
"thinc<8.2.5; python_version<'3.9'",
"thinc>=8.2.5; python_version>='3.9'",
# blis>1.2.0 (dependency of thinc) doesn't provide binaries for python<3.10 so we need to cap it ourself
"blis<1.0.0; python_version<'3.9'",
"blis<1.2.1; python_version>='3.9' and python_version<'3.10'",
"confit>=0.7.3",
"tqdm",
"umls-downloader>=0.1.1",
"numpy>=1.15.0",
"pandas>=1.1.0; python_version<'3.8'",
"pandas>=1.4.0; python_version>='3.8'",
"typing-extensions>=4.0.0",
"dill",
# Packaging
"build>=1.0.0",
"toml",
"xxhash",
"pyarrow>=3.0.0", # support for fragment.metadata.num_rows
"fsspec; python_version>='3.8'",
"fsspec<2023.1.0; python_version<'3.8'",
# this is only to avoid backtracking issues with spacy's capping
"pydantic>=1.10.2",
"pydantic<2.0.0; python_version<'3.8'",
"pydantic-core<2.0.0; python_version<'3.8'",
]
[project.optional-dependencies]
dev-no-ml = [
"pre-commit>=2.0.0; python_version<'3.8'",
"pre-commit>=2.21.0; python_version>='3.8'",
"pytest>=7.1.0",
"pytest-cov>=3.0.0",
"pytest-timeout",
# Data libs
"koalas>=1.8.1; python_version<'3.8'",
"pyspark",
"polars",
"scikit-learn",
# Packaging
"poetry",
"edsnlp[docs-no-ml]",
]
docs-no-ml = [
"mike~=1.1.2",
"mkdocs-charts-plugin==0.0.8",
"mkdocs-img2fig-plugin==0.9.3",
"mkdocs-material~=9.2.0",
"mkdocs-section-index==0.3.4",
"mkdocs~=1.5.2",
"mkdocstrings>=0.20,<0.28.0",
"mkdocstrings-python~=1.1",
"mkdocs-minify-plugin",
"mkdocs-redirects>=1.2.1;python_version>='3.8'",
"pybtex~=0.24.0",
"pathspec>=0.11.1", # required by vendored mkdocs-autorefs PR
"astunparse",
"griffe<0.39",
"jedi",
"html5lib",
]
ml = [
"rich-logger>=0.3.1",
# TODO: uv doesn't seem to resolve torch correctly, unless we cap it ourself
"torch>=1.13.0,<2.0.0; python_version<'3.8'",
"torch>=1.13.0,<2.5.0; python_version<'3.9'",
"torch>=1.13.0; python_version>='3.9'",
"foldedtensor>=0.4.0",
"safetensors>=0.3.0; python_version>='3.8'",
"safetensors>=0.3.0,<0.5.0; python_version<'3.8'",
"transformers>=4.0.0,<5.0.0",
"accelerate>=0.20.3,<1.0.0",
]
docs = [
"edsnlp[docs-no-ml]",
"edsnlp[ml]",
]
dev = [
"edsnlp[dev-no-ml]",
"edsnlp[ml]",
"optuna>=4.0.0",
"plotly>=5.18.0", # required by optuna viz
"ruamel.yaml>=0.18.0",
"configobj>=5.0.9",
"scikit-learn",
]
setup = [
"mlconjug3<3.9.0", # bug https://github.com/Ars-Linguistica/mlconjug3/pull/506
"numpy<2", # mlconjug has scikit-learn dep which doesn't support for numpy 2 yet
]
[project.urls]
"Source Code" = "https://github.com/aphp/edsnlp"
"Documentation" = "https://aphp.github.io/edsnlp"
"Demo" = "https://aphp.github.io/edsnlp/demo"
"Bug Tracker" = "https://github.com/aphp/edsnlp/issues"
[tool.setuptools.dynamic]
version = { attr = "edsnlp.__version__" }
[tool.setuptools.package-data]
"edsnlp" = [
"**/*.pyx",
"**/*.pxd",
"**/*.pxi",
"resources/*.csv",
"resources/*.json",
"resources/*.csv.gz",
"resources/*.json.gz",
]
[tool.setuptools.packages.find]
where = ["."]
[project.entry-points."edsnlp_core"]
"pipeline" = "edsnlp.core.pipeline:Pipeline"
"load" = "edsnlp.core.pipeline:load"
"optimizer" = "edsnlp.training.optimizer:ScheduledOptimizer"
[project.entry-points."spacy_factories"]
# Core
"eds.accents" = "edsnlp.pipes.core.normalizer.accents.factory:create_component"
"eds.contextual_matcher" = "edsnlp.pipes.core.contextual_matcher.factory:create_component"
"eds.endlines" = "edsnlp.pipes.core.endlines.factory:create_component"
"eds.matcher" = "edsnlp.pipes.core.matcher.factory:create_component"
"eds.normalizer" = "edsnlp.pipes.core.normalizer.factory:create_component"
"eds.pollution" = "edsnlp.pipes.core.normalizer.pollution.factory:create_component"
"eds.quotes" = "edsnlp.pipes.core.normalizer.quotes.factory:create_component"
"eds.remove_lowercase" = "edsnlp.pipes.core.normalizer.remove_lowercase.factory:create_component"
"eds.sentences" = "edsnlp.pipes.core.sentences.factory:create_component"
"eds.spaces" = "edsnlp.pipes.core.normalizer.spaces.factory:create_component"
"eds.terminology" = "edsnlp.pipes.core.terminology.factory:create_component"
# NER
"eds.adicap" = "edsnlp.pipes.ner.adicap.factory:create_component"
"eds.emergency_ccmu" = "edsnlp.pipes.ner.scores.emergency.ccmu.factory:create_component"
"eds.charlson" = "edsnlp.pipes.ner.scores.charlson.factory:create_component"
"eds.cim10" = "edsnlp.pipes.ner.cim10.factory:create_component"
"eds.covid" = "edsnlp.pipes.ner.covid.factory:create_component"
"eds.drugs" = "edsnlp.pipes.ner.drugs.factory:create_component"
"eds.elston_ellis" = "edsnlp.pipes.ner.scores.elston_ellis.factory:create_component"
"eds.emergency_gemsa" = "edsnlp.pipes.ner.scores.emergency.gemsa.factory:create_component"
"eds.emergency_priority" = "edsnlp.pipes.ner.scores.emergency.priority.factory:create_component"
"eds.score" = "edsnlp.pipes.ner.scores.factory:create_component"
"eds.sofa" = "edsnlp.pipes.ner.scores.sofa.factory:create_component"
"eds.tnm" = "edsnlp.pipes.ner.tnm.factory:create_component"
"eds.umls" = "edsnlp.pipes.ner.umls.factory:create_component"
"eds.suicide_attempt" = "edsnlp.pipes.ner.suicide_attempt.factory:create_component"
# NER/Comorbidities
"eds.aids" = "edsnlp.pipes.ner.disorders.aids.factory:create_component"
"eds.alcohol" = "edsnlp.pipes.ner.behaviors.alcohol.factory:create_component"
"eds.cerebrovascular_accident" = "edsnlp.pipes.ner.disorders.cerebrovascular_accident.factory:create_component"
"eds.ckd" = "edsnlp.pipes.ner.disorders.ckd.factory:create_component"
"eds.congestive_heart_failure" = "edsnlp.pipes.ner.disorders.congestive_heart_failure.factory:create_component"
"eds.connective_tissue_disease" = "edsnlp.pipes.ner.disorders.connective_tissue_disease.factory:create_component"
"eds.copd" = "edsnlp.pipes.ner.disorders.copd.factory:create_component"
"eds.dementia" = "edsnlp.pipes.ner.disorders.dementia.factory:create_component"
"eds.diabetes" = "edsnlp.pipes.ner.disorders.diabetes.factory:create_component"
"eds.hemiplegia" = "edsnlp.pipes.ner.disorders.hemiplegia.factory:create_component"
"eds.leukemia" = "edsnlp.pipes.ner.disorders.leukemia.factory:create_component"
"eds.liver_disease" = "edsnlp.pipes.ner.disorders.liver_disease.factory:create_component"
"eds.lymphoma" = "edsnlp.pipes.ner.disorders.lymphoma.factory:create_component"
"eds.myocardial_infarction" = "edsnlp.pipes.ner.disorders.myocardial_infarction.factory:create_component"
"eds.peptic_ulcer_disease" = "edsnlp.pipes.ner.disorders.peptic_ulcer_disease.factory:create_component"
"eds.peripheral_vascular_disease" = "edsnlp.pipes.ner.disorders.peripheral_vascular_disease.factory:create_component"
"eds.solid_tumor" = "edsnlp.pipes.ner.disorders.solid_tumor.factory:create_component"
"eds.tobacco" = "edsnlp.pipes.ner.behaviors.tobacco.factory:create_component"
# Span classifiers
"eds.family" = "edsnlp.pipes.qualifiers.family.factory:create_component"
"eds.history" = "edsnlp.pipes.qualifiers.history.factory:create_component"
"eds.hypothesis" = "edsnlp.pipes.qualifiers.hypothesis.factory:create_component"
"eds.negation" = "edsnlp.pipes.qualifiers.negation.factory:create_component"
"eds.reported_speech" = "edsnlp.pipes.qualifiers.reported_speech.factory:create_component"
# Misc
"eds.consultation_dates" = "edsnlp.pipes.misc.consultation_dates.factory:create_component"
"eds.dates" = "edsnlp.pipes.misc.dates.factory:create_component"
"eds.quantities" = "edsnlp.pipes.misc.quantities.factory:create_component"
"eds.reason" = "edsnlp.pipes.misc.reason.factory:create_component"
"eds.sections" = "edsnlp.pipes.misc.sections.factory:create_component"
"eds.tables" = "edsnlp.pipes.misc.tables.factory:create_component"
# Data
"eds.split" = "edsnlp.pipes.misc.split.split:Split"
"eds.standoff_dict2doc" = "edsnlp.data.converters:StandoffDict2DocConverter"
"eds.standoff_doc2dict" = "edsnlp.data.converters:StandoffDoc2DictConverter"
"eds.conll_dict2doc" = "edsnlp.data.converters:ConllDict2DocConverter"
"eds.omop_dict2doc" = "edsnlp.data.converters:OmopDict2DocConverter"
"eds.omop_doc2dict" = "edsnlp.data.converters:OmopDoc2DictConverter"
"eds.ents_doc2dict" = "edsnlp.data.converters:EntsDoc2DictConverter"
# Deprecated (links to the same factories as above)
"SOFA" = "edsnlp.pipes.ner.scores.sofa.factory:create_component"
"accents" = "edsnlp.pipes.core.normalizer.accents.factory:create_component"
"charlson" = "edsnlp.pipes.ner.scores.charlson.factory:create_component"
"consultation_dates" = "edsnlp.pipes.misc.consultation_dates.factory:create_component"
"contextual-matcher" = "edsnlp.pipes.core.contextual_matcher.factory:create_component"
"eds.contextual-matcher" = "edsnlp.pipes.core.contextual_matcher.factory:create_component"
"dates" = "edsnlp.pipes.misc.dates.factory:create_component"
"eds.AIDS" = "edsnlp.pipes.ner.disorders.aids.factory:create_component"
"eds.CKD" = "edsnlp.pipes.ner.disorders.ckd.factory:create_component"
"eds.COPD" = "edsnlp.pipes.ner.disorders.copd.factory:create_component"
"eds.SOFA" = "edsnlp.pipes.ner.scores.sofa.factory:create_component"
"eds.TNM" = "edsnlp.pipes.ner.tnm.factory:create_component"
"eds.elston-ellis" = "edsnlp.pipes.ner.scores.elston_ellis.factory:create_component"
"eds.elstonellis" = "edsnlp.pipes.ner.scores.elston_ellis.factory:create_component"
"eds.emergency.ccmu" = "edsnlp.pipes.ner.scores.emergency.ccmu.factory:create_component"
"eds.emergency.gemsa" = "edsnlp.pipes.ner.scores.emergency.gemsa.factory:create_component"
"eds.emergency.priority" = "edsnlp.pipes.ner.scores.emergency.priority.factory:create_component"
"eds.measures" = "edsnlp.pipes.misc.quantities.factory:create_component"
"eds.measurements" = "edsnlp.pipes.misc.quantities.factory:create_component"
"eds.remove-lowercase" = "edsnlp.pipes.core.normalizer.remove_lowercase.factory:create_component"
"emergency.ccmu" = "edsnlp.pipes.ner.scores.emergency.ccmu.factory:create_component"
"emergency.gemsa" = "edsnlp.pipes.ner.scores.emergency.gemsa.factory:create_component"
"emergency.priority" = "edsnlp.pipes.ner.scores.emergency.priority.factory:create_component"
"endlines" = "edsnlp.pipes.core.endlines.factory:create_component"
"family" = "edsnlp.pipes.qualifiers.family.factory:create_component"
"hypothesis" = "edsnlp.pipes.qualifiers.hypothesis.factory:create_component"
"matcher" = "edsnlp.pipes.core.matcher.factory:create_component"
"negation" = "edsnlp.pipes.qualifiers.negation.factory:create_component"
"normalizer" = "edsnlp.pipes.core.normalizer.factory:create_component"
"pollution" = "edsnlp.pipes.core.normalizer.pollution.factory:create_component"
"quotes" = "edsnlp.pipes.core.normalizer.quotes.factory:create_component"
"reason" = "edsnlp.pipes.misc.reason.factory:create_component"
"remove-lowercase" = "edsnlp.pipes.core.normalizer.remove_lowercase.factory:create_component"
"reported_speech" = "edsnlp.pipes.qualifiers.reported_speech.factory:create_component"
"rspeech" = "edsnlp.pipes.qualifiers.reported_speech.factory:create_component"
"score" = "edsnlp.pipes.ner.scores.factory:create_component"
"sections" = "edsnlp.pipes.misc.sections.factory:create_component"
"sentences" = "edsnlp.pipes.core.sentences.factory:create_component"
"spaces" = "edsnlp.pipes.core.normalizer.spaces.factory:create_component"
"tables" = "edsnlp.pipes.misc.tables.factory:create_component"
"terminology" = "edsnlp.pipes.core.terminology.factory:create_component"
# We could use spacy_factories in principle, but spacy auto-imports all entry points
# by default, when some of ours require optional dependencies (on the other hand,
# edsnlp only import an entrypoint if it is requested). By splitting our pipes between
# spacy_factories and edsnlp_factories, spacy will only look in the dict above and
# edsnlp will look both in the above dict and in the one below.
[project.entry-points."edsnlp_factories"]
# Trainable
"eds.transformer" = "edsnlp.pipes.trainable.embeddings.transformer.factory:create_component"
"eds.text_cnn" = "edsnlp.pipes.trainable.embeddings.text_cnn.factory:create_component"
"eds.span_pooler" = "edsnlp.pipes.trainable.embeddings.span_pooler.factory:create_component"
"eds.ner_crf" = "edsnlp.pipes.trainable.ner_crf.factory:create_component"
"eds.extractive_qa" = "edsnlp.pipes.trainable.extractive_qa.factory:create_component"
"eds.nested_ner" = "edsnlp.pipes.trainable.ner_crf.factory:create_component"
"eds.span_qualifier" = "edsnlp.pipes.trainable.span_classifier.factory:create_component"
"eds.span_classifier" = "edsnlp.pipes.trainable.span_classifier.factory:create_component"
"eds.span_linker" = "edsnlp.pipes.trainable.span_linker.factory:create_component"
"eds.biaffine_dep_parser" = "edsnlp.pipes.trainable.biaffine_dep_parser.factory:create_component"
[project.entry-points."spacy_scorers"]
"eds.ner_exact" = "edsnlp.metrics.ner:NerExactMetric"
"eds.ner_token" = "edsnlp.metrics.ner:NerTokenMetric"
"eds.ner_overlap" = "edsnlp.metrics.ner:NerOverlapMetric"
"eds.span_attributes" = "edsnlp.metrics.span_attributes:SpanAttributeMetric"
"eds.dep_parsing" = "edsnlp.metrics.dep_parsing:DependencyParsingMetric"
# Deprecated
"eds.ner_exact_metric" = "edsnlp.metrics.ner:NerExactMetric"
"eds.ner_token_metric" = "edsnlp.metrics.ner:NerTokenMetric"
"eds.ner_overlap_metric" = "edsnlp.metrics.ner:NerOverlapMetric"
"eds.span_attributes_metric" = "edsnlp.metrics.span_attributes:SpanAttributeMetric"
"eds.ner_exact_scorer" = "edsnlp.metrics.ner:NerExactMetric"
"eds.ner_token_scorer" = "edsnlp.metrics.ner:NerTokenMetric"
"eds.ner_overlap_scorer" = "edsnlp.metrics.ner:NerOverlapMetric"
"eds.span_attributes_scorer" = "edsnlp.metrics.span_attributes:SpanAttributeMetric"
[project.entry-points."edsnlp_readers"]
"spark" = "edsnlp.data:from_spark"
"pandas" = "edsnlp.data:from_pandas"
"json" = "edsnlp.data:read_json"
"parquet" = "edsnlp.data:read_parquet"
"standoff" = "edsnlp.data:read_standoff"
"brat" = "edsnlp.data:read_brat" # alias for standoff
"conll" = "edsnlp.data:read_conll"
[project.entry-points."edsnlp_writers"]
"spark" = "edsnlp.data:to_spark"
"pandas" = "edsnlp.data:to_pandas"
"json" = "edsnlp.data:write_json"
"standoff" = "edsnlp.data:write_standoff"
"brat" = "edsnlp.data:write_brat" # alias for standoff
[project.entry-points."spacy_misc"]
"eds.span_context_getter" = "edsnlp.utils.span_getters:make_span_context_getter"
[project.entry-points."spacy_languages"]
"eds" = "edsnlp.language:EDSLanguage"
[project.entry-points."spacy_tokenizers"]
"eds.tokenizer" = "edsnlp.language:create_eds_tokenizer"
[project.entry-points."mkdocs.plugins"]
"bibtex" = "docs.scripts.bibtex:BibTexPlugin"
"autorefs" = "docs.scripts.autorefs.plugin:AutorefsPlugin"
"clickable_snippets" = "docs.scripts.clickable_snippets:ClickableSnippetsPlugin"
[build-system]
requires = [
"setuptools",
"cython>=0.25",
"spacy>=3.2,!=3.8.2; python_version<'3.9'",
"spacy>=3.2,!=3.8.2,<4.0.0; python_version>='3.9'",
# thinc doesn't provide binaries for python<3.9 from 8.2.5 so we need to cap it ourselves
"thinc<8.2.5; python_version<'3.9'",
"thinc>=8.2.5; python_version>='3.9'",
# to update from https://github.com/scipy/oldest-supported-numpy/blob/main/setup.cfg
# while setting numpy >= 1.15.0 due to spacy reqs
"numpy==1.15.0; python_version=='3.7' and platform_machine not in 'arm64|aarch64|loongarch64' and platform_system!='AIX' and platform_python_implementation != 'PyPy'",
"numpy==1.15.0; python_version=='3.7' and platform_machine=='arm64' and platform_system=='Windows' and platform_python_implementation != 'PyPy'",
"numpy==1.16.0; python_version=='3.7' and platform_system=='AIX' and platform_machine!='loongarch64' and platform_python_implementation != 'PyPy'",
"numpy==1.17.3; python_version=='3.8' and platform_machine not in 'arm64|aarch64|s390x|loongarch64' and platform_python_implementation != 'PyPy'",
"numpy==1.17.3; python_version=='3.8' and platform_machine=='arm64' and platform_system=='Windows' and platform_python_implementation != 'PyPy'",
"numpy==1.17.5; python_version=='3.8' and platform_machine=='s390x' and platform_python_implementation != 'PyPy'",
"numpy==1.19.0; python_version=='3.6' and platform_machine!='loongarch64' and platform_python_implementation=='PyPy'",
"numpy==1.19.2; python_version=='3.7' and platform_machine=='aarch64' and platform_system!='AIX' and platform_python_implementation != 'PyPy'",
"numpy==1.19.2; python_version=='3.8' and platform_machine=='aarch64' and platform_python_implementation != 'PyPy'",
"numpy==1.20.0; python_version=='3.7' and platform_machine!='loongarch64' and platform_python_implementation=='PyPy'",
"numpy==1.21.0; python_version=='3.7' and platform_machine=='arm64' and platform_system=='Darwin' and platform_python_implementation!='PyPy'",
"numpy==1.21.0; python_version=='3.8' and platform_machine=='arm64' and platform_system=='Darwin' and platform_python_implementation!='PyPy'",
"numpy==1.22.2; python_version>='3.8' and python_version<'3.9' and platform_machine=='loongarch64' and platform_python_implementation!='PyPy'",
"numpy==1.22.2; python_version=='3.8' and platform_machine!='loongarch64' and platform_python_implementation=='PyPy'",
"numpy>=2.0; python_version>='3.9'",
"blis<1.0.0; python_version<'3.9'",
"blis<1.2.1; python_version>='3.9' and python_version<'3.10'",
]
build-backend = "setuptools.build_meta"
[tool.ruff]
fix = true
extend-exclude = [
".git",
"__pycache__",
"__init__.py",
".mypy_cache",
".pytest_cache",
".venv",
"build",
"edsnlp/pipes/factories.py",
]
line-length = 88
[tool.ruff.lint]
select = [
"E",
"F",
"W",
"I001"
]
[tool.ruff.lint.flake8-tidy-imports]
ban-relative-imports = "parents"
[tool.ruff.lint.extend-per-file-ignores]
"__init__.py" = ["F401"]
"edsnlp/pipes/factories.py" = [ "F401", "E501" ]
[tool.ruff.lint.isort]
known-first-party = ["edsnlp"]
known-third-party = ["build"]
order-by-type = true
[tool.interrogate]
ignore-init-method = true
ignore-init-module = true
ignore-magic = false
ignore-semiprivate = true
ignore-private = true
ignore-property-decorators = true
ignore-module = true
ignore-nested-functions = true
ignore-nested-classes = true
ignore-setters = true
fail-under = 40
exclude = ["setup.py", "docs", "build", "tests", "edsnlp/pipes/core/contextual_matcher/models.py"]
verbose = 0
quiet = false
whitelist-regex = []
ignore-regex = ['__(?!init).*__']
color = true
omit-covered-files = false
# generate-badge = "."
# badge-format = "svg"
[tool.pytest.ini_options]
# Some tests may download large objects such as the UMLS
# This timeout is mostly here to kill the CI in the case of deadlocks or infinite loops
timeout = 600
[tool.coverage.report]
precision = 2
include = ["edsnlp/*"]
omit = [
"tests/*",
]
exclude_lines = [
"def __repr__",
"if __name__ == .__main__.:",
"@overload",
"pragma: no cover",
"raise .*Error",
"raise .*Exception",
"warn\\(",
"if __name__ == .__main__.:",
"if repr_id in exclude:",
"if TYPE_CHECKING:",
"class .*\\bProtocol\\):",
"@(abc\\.)?abstractmethod",
"Span.set_extension.*",
"Doc.set_extension.*",
"Token.set_extension.*",
]
[tool.coverage.run]
include = ["edsnlp/*"]
concurrency = ["multiprocessing", "thread"]
parallel = true
[tool.uv.pip]
torch-backend = "auto"
[tool.cibuildwheel]
skip = [
"*p36-*", # Skip Python 3.6
"pp*", # Skip PyPy
"*-win32", # Skip 32-bit Windows
"*-manylinux_i686", # Skip 32-bit Linux
"*-win_arm64", # Skip experimental Windows on ARM
"*-musllinux*", # Skip slow Linux
"*-manylinux_ppc64le", # Skip slow Linux
"*-manylinux_s390x", # Skip slow Linux
"*p313-*", # skip 3.13 for now (spacy build issue): users will have to build it on their own
]
before-test = 'pip install pytest "urllib3<2"'
test-extras = "ml"
test-command = "pytest {project}/tests/pipelines/test_pipelines.py"
