pyproject.toml

# pyproject.toml — MyWebIntelligence (MWI)
#
# Source of truth for: project metadata, runtime/ML/dev dependencies, and mypy.
# Deliberately NOT here, by project convention:
#   - lint config              → .flake8       (flake8 can't read pyproject)
#   - pytest + coverage config → pytest.ini    (.claude/rules/testing.md mandates it)
#   - pinned installs          → uv.lock + the generated requirements.txt (`make lock`)
#
# MWI is a flat-layout CLI app (`python mywi.py`, PYTHONPATH=.), not an installable
# library: [tool.uv] package = false, no [build-system], no [project.scripts].

[project]
name = "mywebintelligence"
version = "1.0.0"
description = "Reproducible web-mining toolkit for digital methods in the social sciences and humanities (MyWebIntelligence / MWI)."
readme = "README.md"
license = { file = "LICENSE" }
# Support floor. Mirror it in mypy `python_version` below and the CI matrix
# (.github/workflows/tests.yml). The pinned dev/runtime interpreter is 3.11
# (.python-version, Dockerfile); 3.9 is only the lowest version we still support.
requires-python = ">=3.9"
authors = [
    { name = "Amar Lakel", email = "amar.lakel@u-bordeaux-montaigne.fr" },
]

# Runtime dependencies — first-order imports only (what mwi/ actually imports).
# Pure transitive deps (aiosignal, async-timeout, attrs, frozenlist, multidict,
# yarl, soupsieve, joblib, urllib3, idna, wrapt) are resolved and pinned by
# uv.lock; listing them here just freezes a pip-freeze snapshot and hides intent.
# requirements.txt is a generated artifact (`make lock` → `uv export`).
#
# NOTE: this array is a bare key of [project] and MUST stay above any
# [project.*] subtable (urls / optional-dependencies) — TOML would otherwise
# attach it to that subtable.
dependencies = [
    # Upper bound: aiohttp 3.14 made ClientResponse.__init__ require a new
    # `stream_writer` kwarg that the latest aioresponses (0.7.8) test mock does
    # not pass — breaking the search-provider tests. Production paths work on
    # 3.13.x. Lift this cap once aioresponses ships aiohttp-3.14 support.
    "aiohttp>=3.9.3,<3.14",
    "beautifulsoup4>=4.12.3",
    "certifi>=2023.1.1",     # imported directly in core.py for SSL cert hardening
    "colorama>=0.4.4",       # optional colored output in scripts/install_utils.py
    "curl_cffi>=0.7.0",      # TLS impersonation, sprint-403 fetch cascade
    "langdetect>=1.0.9",
    "lxml>=4.9.0",
    "nltk>=3.8",
    "numpy>=1.26.4",
    "peewee>=3.15.0",
    "Pillow>=10.0.0",
    "playwright>=1.42.0",    # browser fallback + dynamic media extraction
    "requests>=2.31.0",
    "scikit-learn>=1.4.2",
    "trafilatura>=1.6.0",
]

[project.urls]
Repository = "https://github.com/MyWebIntelligence/My-Web-Intelligence-v2"
Issues = "https://github.com/MyWebIntelligence/My-Web-Intelligence-v2/issues"

[project.optional-dependencies]
# Optional ML extras (ANN + NLI). Every one is used — conditionally imported in
# mwi/semantic_pipeline.py, with a neutral fallback when absent. Install with
# `uv sync --extra ml` (mirrors the hand-kept requirements-ml.txt).
ml = [
    "torch>=2.2",
    "transformers>=4.40",
    "tokenizers>=0.15",
    "sentencepiece>=0.1.99",
    "sentence-transformers>=2.6",
    "faiss-cpu>=1.12.0",
]

# Development / test tooling (PEP 735). Installed by default on `uv sync`,
# skipped with `--no-dev`. Lint = flake8 (config in .flake8); types = mypy
# (config below). pytest-asyncio + aioresponses back the search-router tests.
[dependency-groups]
dev = [
    "pytest>=7.0.0",
    "pytest-cov>=4.0.0",
    "pytest-asyncio>=0.23.0",
    "aioresponses>=0.7.6",
    "flake8>=6.0.0",
    "mypy>=1.0.0",
]

[tool.uv]
# MWI is an application run via `python mywi.py` (flat layout, PYTHONPATH=.),
# not an installable library. `package = false` installs the dependencies into
# the environment without building/installing the project itself — hence no
# [build-system] and no [project.scripts].
package = false

# ===========================================================================
# MYPY — types. Progressive adoption: no global strict (legacy code is only
# partially typed and Peewee resists static typing). Strict tightens per-module.
# Run on touched code: `uv run mypy mwi/`. Lint config lives in .flake8.
# ===========================================================================
[tool.mypy]
python_version = "3.9"        # check against the support floor, not the dev 3.11
show_error_codes = true
warn_redundant_casts = true
warn_unused_ignores = true
warn_unused_configs = true
no_implicit_optional = true
check_untyped_defs = true      # check bodies of unannotated funcs without requiring annotations

[[tool.mypy.overrides]]
# Recent, fully-typed modules: require strict typing here. Extend this list as
# more modules get typed.
module = ["mwi.search.*"]
disallow_untyped_defs = true
warn_return_any = true

[[tool.mypy.overrides]]
# Third-party deps without type stubs: silence missing-import errors without
# disabling mypy globally.
module = [
    "peewee.*",
    "trafilatura.*",
    "langdetect.*",
    "curl_cffi.*",
    "nltk.*",
    "sklearn.*",
    "faiss.*",
    "sentence_transformers.*",
    "transformers.*",
]
ignore_missing_imports = true