poly.toml

[discovery]
exclude = [
    "**/*.freezed.dart",
    "**/*.g.dart",
    "**/*.jinja",
    "**/*.lock",
    "**/Cargo.lock",
    "**/go.sum",
    "**/package-lock.json",
    "**/pnpm-lock.yaml",
    "**/uv.lock",
    ".alef/**",
    "artifacts/**",
    "dist/**",
    "docs/assets/**",
    "docs/snippets/**",
    "fixtures/**",
    "node_modules/**",
    "readme_templates/**",
    "target/**",
    "templates/readme/**",
    "test_documents/**",
    "vendor/**",
    "**/vendored/**",
    "**/stopwords/**",
    "**/testdata/**",
    "e2e/**",
    "**/*.diff",
    "**/*.patch",
    "**/_generated/**",
    "**/coverage/**",
    "**/pkg/**",
    "**/.next/**",
    "**/*.min.js",
    "**/frb_generated.rs",
]

[lint.markdown.rumdl]
disable = ["MD012", "MD013", "MD024", "MD033", "MD036", "MD041", "MD046", "MD051", "MD076"]

[fmt.markdown.rumdl]
disable = ["MD012", "MD013", "MD024", "MD033", "MD036", "MD041", "MD046", "MD051", "MD076"]

[lint.python.ruff]
select = ["ALL"]
ignore = [
    "ANN401",
    "ASYNC109",
    "ASYNC110",
    "BLE001",
    "COM812",
    "D100",
    "D104",
    "D107",
    "D205",
    "E501",
    "EM",
    "FBT",
    "FIX",
    "ISC001",
    "PD011",
    "PGH003",
    "PLR2004",
    "PLW0603",
    "S104",
    "S110",
    "S603",
    "TD",
    "TRY",
]
mccabe_max_complexity = 15
pydocstyle_convention = "google"
pylint_max_args = 10
pylint_max_branches = 15
pylint_max_returns = 10

[lint.php.mago]
select = ["correctness", "security"]
ignore = ["strict-assertions", "use-specific-assertions", "no-redundant-variable", "sensitive-parameter"]
php_version = "8.2"


# Spell-checker allowlist. Data/corpus trees (testdata tokenizer vocab, stopword
# lists, vendored benchmark ground-truth) are handled by [discovery] exclude
# above; these entries whitelist valid identifiers/terms in real source so autofix
# never mangles them (the original bug: the `flate2` crate got "corrected" to `flat2`).
[lint.typos.extend_words]
# flate2 crate + PDF FlateDecode filter (THE ORIGINAL BUG: flate -> flat)
flate = "flate"
# macOS wheel-vendoring tool (publish.yaml)
delocate = "delocate"
# candle-ocr ML tensor/API terms
thw = "thw"       # image_grid_thw / grid_thw tensor dim
arange = "arange" # candle Tensor::arange (covers "Arange")
mmaped = "mmaped" # candle from_mmaped_safetensors
# binding / keyword-algorithm split tokens
ser = "ser"               # Jackson databind.ser.std (Java/Kotlin bindings)
ake = "ake"               # yAKE keyword algorithm (go binding)
algoritmos = "algoritmos" # Spanish test string (keywords_integration.rs)
# Project / product / tool names
kreuzberg = "kreuzberg"
zensical = "zensical"
tesseract = "tesseract"
onnx = "onnx"
surrealdb = "surrealdb"
docling = "docling"
markitdown = "markitdown"
pymupdf = "pymupdf"
openwebui = "openwebui"
webui = "webui"
mkdocs = "mkdocs"
mkdocstrings = "mkdocstrings"
rumdl = "rumdl"
# Document-format domain terms
opf = "opf"           # EPUB Open Packaging Format
hocr = "hocr"
odf = "odf"
multline = "multline" # LaTeX environment
headerr = "headerr"   # RTF control word
pard = "pard"         # RTF control word
edn = "edn"           # DOC field name
siz = "siz"           # JPEG2000 SIZ marker
# ISO-639 / OCR language codes
inh = "inh"
bre = "bre"
yor = "yor"
# Upstream (Tesseract / PDFium / PaddleOCR) API/constant spellings
extention = "extention"
portait = "portait"
fith = "fith"
threed = "threed"
chlidren = "chlidren"
formated = "formated"
specifing = "specifing"
similarily = "similarily"
execpt = "execpt"
faiure = "faiure"
substract = "substract"
charater = "charater"
cliper = "cliper"
splitted = "splitted"
unparseable = "unparseable"
relatie = "relatie"         # Dutch "relatie-id" test placeholder
traineddata = "traineddata"
# Language codes / foreign-language test strings in real source
tha = "tha"                     # Thai ISO-639 code (OCR language lists)
ist = "ist"                     # German "ist" in language-detection samples
internacional = "internacional" # es/pt language-detection sample
utiliza = "utiliza"             # Spanish keyword-extraction test string
# Short domain fragments flagged in real source
caf = "caf"
hom = "hom"   # LaTeX \hom operator
ths = "ths"   # DOCX 240ths measurement unit
ment = "ment" # semantic token-reduction fragment
mis = "mis"
helo = "helo"
ove = "ove"   # OVE fragment in pdf_text_merging test
tre = "tre"   # Tre fragment (pdfium-render)
thr = "thr"   # thr fragment (benchmark-harness aggregate)
# Foreign-language words in test strings / translation guide
programa = "programa"
cursos = "cursos"
excepcional = "excepcional"
ein = "ein"                 # German "ein"
legt = "legt"               # German "legt" (translation.md)

[lint.typos.extend_identifiers]
PyMuPDF = "PyMuPDF"
MarkItDown = "MarkItDown"
SurrealDB = "SurrealDB"
PDFium = "PDFium"
WebUI = "WebUI"
traineddata = "traineddata"

[per-file-ignores]
# text_repair.rs contains intentionally broken/split words as test fixtures
# (hyphenation-rejoin cases) — ignore typos there rather than globally
# allowlisting real-typo-looking fragments.
"**/text_repair.rs" = ["typo"]
# alef/napi-generated JS loader + wrapper — unused-var nits are generator output,
# not hand-editable here (fix belongs in alef's napi backend).
"crates/xberg-node/index.js" = ["no-unused-vars"]
"crates/xberg-node/format-metadata-wrapper.js" = ["no-unused-vars"]
# Dockerfiles: DL3006 fires only on unpinnable `FROM scratch` (false positive);
# DL3059 (consecutive RUN) is stylistic — multi-stage RUNs kept split on purpose.
# (The dockerfile engine ignores [lint.*] config, so suppress by path here.)
"docker/**" = ["DL3006", "DL3059"]
"**/api.py" = ["F401", "I001", "TC006", "UP035"]
"**/*.pyi" = ["A002", "F401", "I001", "PYI033", "TC006", "UP035"]
"**/options.py" = ["F401", "I001", "RUF100"]
"**/__init__.py" = ["I001"]
"**/tests/**" = [
    "ANN",
    "D103",
    "PLR2004",
    "PLR0915",
    "PLR0913",
    "S101",
    "S105",
    "S106",
    "S108",
    "S310",
    "S311",
    "PT011",
    "PT012",
    "PERF401",
    "PTH123",
    "T201",
    "TC001",
    "TC002",
    "TC003",
    "INP001",
    "no-unused-vars",
    "no-literal-password",
    "no-unescaped-output",
    "RUF100",
    "F401",
    "F811",
    "I001",
    "PT018",
    "ARG001",
    "ARG002",
    "D403",
    "E713",
    "UP035",
    "UP012",
    "RUF015",
    "F541",
    "EXE001",
    "A001",
    "N801",
]
"**/e2e/**" = [
    "ANN",
    "D103",
    "PLR2004",
    "PLR0915",
    "PLR0913",
    "S101",
    "S105",
    "S106",
    "S108",
    "S310",
    "S311",
    "PT011",
    "PT012",
    "PERF401",
    "PTH123",
    "T201",
    "TC001",
    "TC002",
    "TC003",
    "INP001",
    "no-unused-vars",
    "no-literal-password",
    "no-unescaped-output",
    "RUF100",
    "F401",
    "F811",
    "I001",
    "PT018",
    "ARG001",
    "ARG002",
    "D403",
    "E713",
    "UP035",
    "UP012",
    "RUF015",
    "F541",
    "EXE001",
    "A001",
    "N801",
]
"**/test_apps/**" = [
    "ANN",
    "D103",
    "PLR2004",
    "PLR0915",
    "PLR0913",
    "S101",
    "S105",
    "S106",
    "S108",
    "S310",
    "S311",
    "PT011",
    "PT012",
    "PERF401",
    "PTH123",
    "T201",
    "TC001",
    "TC002",
    "TC003",
    "INP001",
    "no-unused-vars",
    "no-literal-password",
    "no-unescaped-output",
    "RUF100",
    "F401",
    "F811",
    "I001",
    "PT018",
    "ARG001",
    "ARG002",
    "D403",
    "E713",
    "UP035",
    "UP012",
    "RUF015",
    "F541",
    "EXE001",
    "A001",
    "N801",
]
"cli-proxy/pypi/**/*.py" = ["ANN001", "ANN202", "S310", "T201"]
"e2e/**/*.py" = ["A001", "N801"]
"packages/python/build.py" = ["INP001"]
"packages/python/xberg/options.py" = ["I001"]
"scripts/**/*.py" = [
    "B023",
    "C901",
    "D101",
    "D102",
    "D103",
    "D415",
    "E702",
    "ERA001",
    "PERF203",
    "PERF401",
    "PLC0415",
    "PLR0912",
    "PLR0915",
    "PLW2901",
    "PT018",
    "PTH108",
    "PTH123",
    "S101",
    "S108",
    "S311",
    "S607",
    "SIM105",
    "SIM115",
    "T201",
]
"tools/benchmark-harness/python_baselines/*.py" = [
    "ANN001",
    "ANN201",
    "ANN202",
    "ARG001",
    "BLE001",
    "C901",
    "D103",
    "G004",
    "INP001",
    "PERF401",
    "PLC0415",
    "PLR0911",
    "PLR0912",
    "PLR0915",
    "PLW2901",
    "PTH123",
    "S607",
    "T201",
]
"tools/benchmark-harness/scripts/*.php" = ["no-unescaped-output"]
"tools/benchmark-harness/scripts/*.py" = [
    "ANN001",
    "ANN201",
    "ANN202",
    "ARG001",
    "BLE001",
    "C901",
    "D103",
    "E741",
    "EXE001",
    "EXE002",
    "F841",
    "INP001",
    "N818",
    "PERF203",
    "PERF401",
    "PIE810",
    "PLC0206",
    "PLC0415",
    "PLR0911",
    "PLR0912",
    "PLR0915",
    "PLR1714",
    "PLW1510",
    "PLW2901",
    "PTH110",
    "PTH112",
    "PTH118",
    "PTH119",
    "PTH122",
    "PTH123",
    "RET504",
    "S314",
    "S607",
    "SIM108",
    "SIM115",
    "T201",
]
"tools/generate_test_fixtures/**/*.py" = [
    "D301",
    "ERA001",
    "PERF401",
    "PLC0415",
    "PT001",
    "S101",
    "SIM117",
    "T201",
    "TC003",
]
"tools/perf/*.py" = ["D103", "T201"]

[hooks]
stages = ["pre-commit"]

[hooks.builtin]
polylint = { exclude = [
    "**/*.freezed.dart",
    "**/*.g.dart",
    "**/*.jinja",
    "**/*.lock",
    "**/Cargo.lock",
    "**/go.sum",
    "**/package-lock.json",
    "**/pnpm-lock.yaml",
    "**/uv.lock",
    ".alef/**",
    "artifacts/**",
    "dist/**",
    "docs/assets/**",
    "docs/snippets/**",
    "fixtures/**",
    "node_modules/**",
    "readme_templates/**",
    "target/**",
    "templates/readme/**",
    "test_documents/**",
    "vendor/**",
    "**/vendored/**",
    "**/stopwords/**",
    "**/testdata/**",
    "e2e/**",
    "**/*.diff",
    "**/*.patch",
    "**/_generated/**",
    "**/coverage/**",
    "**/pkg/**",
    "**/.next/**",
    "**/*.min.js",
    "**/frb_generated.rs",
] }
polyfmt = { exclude = [
    "**/*.freezed.dart",
    "**/*.g.dart",
    "**/*.jinja",
    "**/*.lock",
    "**/Cargo.lock",
    "**/go.sum",
    "**/package-lock.json",
    "**/pnpm-lock.yaml",
    "**/uv.lock",
    ".alef/**",
    "artifacts/**",
    "dist/**",
    "docs/assets/**",
    "docs/snippets/**",
    "fixtures/**",
    "node_modules/**",
    "readme_templates/**",
    "target/**",
    "templates/readme/**",
    "test_documents/**",
    "vendor/**",
    "**/vendored/**",
    "**/stopwords/**",
    "**/testdata/**",
    "e2e/**",
    "**/*.diff",
    "**/*.patch",
    "**/_generated/**",
    "**/coverage/**",
    "**/pkg/**",
    "**/.next/**",
    "**/*.min.js",
    "**/frb_generated.rs",
] }
file_safety = { exclude = [
    "**/*.freezed.dart",
    "**/*.g.dart",
    "**/*.jinja",
    "**/*.lock",
    "**/Cargo.lock",
    "**/go.sum",
    "**/package-lock.json",
    "**/pnpm-lock.yaml",
    "**/uv.lock",
    ".alef/**",
    "artifacts/**",
    "dist/**",
    "docs/assets/**",
    "docs/snippets/**",
    "fixtures/**",
    "node_modules/**",
    "readme_templates/**",
    "target/**",
    "templates/readme/**",
    "test_documents/**",
    "vendor/**",
    "**/vendored/**",
    "**/stopwords/**",
    "**/testdata/**",
    "e2e/**",
    "**/*.diff",
    "**/*.patch",
    "**/_generated/**",
    "**/coverage/**",
    "**/pkg/**",
    "**/.next/**",
    "**/*.min.js",
    "**/frb_generated.rs",
] }
cargo = true
commit = { stages = ["commit-msg"] }

[hooks.pre-commit.commands.pyrefly]
run = "pyrefly check packages/python"
files = "packages/python/**/*.py"