sift-lint 1.0.0__tar.gz

sift_lint-1.0.0/.gitattributes

@@ -0,0 +1,3 @@
+# This fixture intentionally mixes CRLF/LF and contains trailing whitespace.
+# Treat it as data so Git never normalizes the bytes on checkout or commit.
+examples/orders_messy.csv -text

sift_lint-1.0.0/.github/ISSUE_TEMPLATE/bug_report.md

@@ -0,0 +1,23 @@
+---
+name: Bug report
+about: Report an incorrect finding, crash, or other Sift problem
+---
+
+<!-- For a false positive — Sift reporting something that is actually fine —
+     please include the smallest CSV that reproduces it. That case usually
+     becomes a permanent regression test. -->
+
+**What Sift reported**
+
+```
+paste the finding here
+```
+
+**What the data actually is**
+
+**Smallest file that reproduces it**
+
+```csv
+```
+
+**Version** — output of `sift-lint --version`

sift_lint-1.0.0/.github/dependabot.yml

@@ -0,0 +1,7 @@
+version: 2
+updates:
+  - package-ecosystem: github-actions
+    directory: "/"
+    schedule:
+      interval: weekly
+    open-pull-requests-limit: 5

sift_lint-1.0.0/.github/scripts/check_distributions.py

@@ -0,0 +1,93 @@
+"""Check built distributions in clean environments outside the source tree."""
+
+from __future__ import annotations
+
+import os
+import pathlib
+import subprocess
+import sys
+import tarfile
+import tempfile
+import venv
+import zipfile
+
+ROOT = pathlib.Path(__file__).resolve().parents[2]
+DIST = ROOT / "dist"
+EXAMPLE = ROOT / "examples" / "orders_messy.csv"
+
+
+def run(*args: str, cwd: pathlib.Path) -> None:
+    subprocess.run(args, cwd=cwd, check=True)
+
+
+def single(pattern: str) -> pathlib.Path:
+    matches = sorted(DIST.glob(pattern))
+    if len(matches) != 1:
+        raise RuntimeError(f"Expected one {pattern} in {DIST}, found {len(matches)}")
+    return matches[0]
+
+
+def check_contents(wheel: pathlib.Path, sdist: pathlib.Path) -> None:
+    with zipfile.ZipFile(wheel) as archive:
+        members = set(archive.namelist())
+        if "sift_lint/__init__.py" not in members or "sift_lint/cli.py" not in members:
+            raise RuntimeError("Wheel is missing the Sift package")
+        if any(name.startswith("sift/") for name in members):
+            raise RuntimeError("Wheel contains the old import package")
+
+    with tarfile.open(sdist, "r:gz") as archive:
+        members = {name.split("/", 1)[-1] for name in archive.getnames()}
+        for expected in ("sift_lint/__init__.py", "README.md", "LICENSE"):
+            if expected not in members:
+                raise RuntimeError(f"Source distribution is missing {expected}")
+
+
+def smoke(artifact: pathlib.Path, workspace: pathlib.Path, label: str) -> None:
+    env = workspace / label
+    venv.EnvBuilder(with_pip=True).create(env)
+    scripts = env / ("Scripts" if os.name == "nt" else "bin")
+    python = scripts / ("python.exe" if os.name == "nt" else "python")
+    executable = scripts / ("sift-lint.exe" if os.name == "nt" else "sift-lint")
+
+    run(str(python), "-m", "pip", "install", "--no-deps", str(artifact), cwd=workspace)
+    run(
+        str(python),
+        "-c",
+        (
+            "from importlib.metadata import version; "
+            "from pathlib import Path; "
+            "import sift_lint, sys; "
+            "assert sift_lint.__version__ == version('sift-lint'); "
+            "assert not Path(sift_lint.__file__).resolve().is_relative_to(Path(sys.argv[1]))"
+        ),
+        str(ROOT),
+        cwd=workspace,
+    )
+    run(str(executable), "--version", cwd=workspace)
+    run(str(executable), "--help", cwd=workspace)
+    run(
+        str(executable),
+        "check",
+        str(EXAMPLE),
+        "--no-config",
+        "--fail-on",
+        "none",
+        cwd=workspace,
+    )
+    run(str(python), "-m", "pip", "check", cwd=workspace)
+
+
+def main() -> None:
+    wheel = single("*.whl")
+    sdist = single("*.tar.gz")
+    check_contents(wheel, sdist)
+    run(sys.executable, "-m", "twine", "check", "--strict", str(wheel), str(sdist), cwd=ROOT)
+
+    with tempfile.TemporaryDirectory(prefix="sift-package-") as directory:
+        workspace = pathlib.Path(directory)
+        smoke(wheel, workspace, "wheel")
+        smoke(sdist, workspace, "sdist")
+
+
+if __name__ == "__main__":
+    main()

sift_lint-1.0.0/.github/scripts/check_release.py

@@ -0,0 +1,33 @@
+"""Reject release tags that do not match the package version."""
+
+from __future__ import annotations
+
+import re
+import sys
+
+from sift_lint import __version__
+
+TAG_PATTERN = re.compile(r"v(0|[1-9][0-9]*)\.(0|[1-9][0-9]*)\.(0|[1-9][0-9]*)\Z")
+
+
+def check_tag(tag: str, version: str) -> None:
+    if not TAG_PATTERN.fullmatch(tag):
+        raise ValueError(f"Expected a stable release tag like v1.0.0, got {tag!r}")
+    if int(tag[1:].split(".", 1)[0]) == 0:
+        raise ValueError("Public releases start at version 1.0.0")
+    if tag[1:] != version:
+        raise ValueError(f"Release tag {tag!r} does not match package version {version!r}")
+
+
+def main() -> None:
+    if len(sys.argv) != 2:
+        raise SystemExit("Usage: python .github/scripts/check_release.py vX.Y.Z")
+    try:
+        check_tag(sys.argv[1], __version__)
+    except ValueError as exc:
+        raise SystemExit(str(exc)) from exc
+    print(f"Release tag {sys.argv[1]} matches package version {__version__}")
+
+
+if __name__ == "__main__":
+    main()

sift_lint-1.0.0/.github/workflows/README.md

@@ -0,0 +1,46 @@
+# Workflows
+
+`ci.yml` runs on every push to `main` and every pull request.
+
+**test** — runs Ruff, the full test suite and the recall benchmark on Ubuntu
+with Python 3.11–3.14, plus Windows and macOS with Python 3.14. The format
+extras are installed for every test job. Python 3.15 will be added when
+`actions/setup-python` provides it for the selected runner.
+
+**package** — builds a wheel and source distribution on Linux, Windows and
+macOS. Both archives undergo strict metadata checks and are installed
+separately in clean virtual environments. Each installed package is tested
+outside the repository for version consistency, imports, the CLI, a real CSV
+check and `pip check`.
+
+**data-gate** — runs Sift against this repository's own example files, the way
+it is meant to be used in a pipeline: `check` on a folder, `diff` against a
+baseline, `fix` as a dry run, and `impact` on a known-bad column. These use
+`--fail-on none` because the example data is deliberately broken; in a real
+pipeline you would drop that flag and let a bad file fail the build.
+
+## Supply-chain controls
+
+Workflow actions are pinned to full commit SHAs rather than mutable version
+tags. Checkout does not persist the repository token because these jobs never
+push back to the repository, and the workflow token is explicitly limited to
+read-only repository contents.
+
+## Release publishing
+
+`release.yml` runs only when a GitHub Release is published. It ignores prereleases
+and rejects tags that do not match the package's stable version or point outside
+`main`. The unprivileged build job runs tests and the benchmark, then builds and
+validates a wheel and source distribution. The separate publish job downloads
+those artifacts and publishes through PyPI Trusted Publishing.
+
+The publish job uses the `pypi` GitHub environment and has only the OIDC
+permission needed for publishing. No long-lived PyPI token is stored. Creating
+a branch, pushing a tag, or merging a pull request never triggers publishing.
+
+Before using **Publish release**, configure the `pypi` environment and the
+matching Trusted Publisher on PyPI, then complete the maintainer's release
+checklist. Publishing a release is the deliberate production publishing action.
+
+`.github/dependabot.yml` proposes weekly updates for GitHub Actions. Review the
+updated commit SHA pins before merging those pull requests.

sift_lint-1.0.0/.github/workflows/ci.yml

@@ -0,0 +1,83 @@
+name: ci
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+
+permissions:
+  contents: read
+
+jobs:
+  test:
+    name: test (${{ matrix.os }}, Python ${{ matrix.python-version }})
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - os: ubuntu-24.04
+            python-version: "3.11"
+          - os: ubuntu-24.04
+            python-version: "3.12"
+          - os: ubuntu-24.04
+            python-version: "3.13"
+          - os: ubuntu-24.04
+            python-version: "3.14"
+          - os: windows-2025
+            python-version: "3.14"
+          - os: macos-15
+            python-version: "3.14"
+    steps:
+      - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
+        with:
+          persist-credentials: false
+      - uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: ${{ matrix.python-version }}
+      - run: python -m pip install -e ".[dev]"
+      - run: ruff check .
+      - run: pytest -q
+      # Preserve the benchmark's corruption-detection and clean-control gates.
+      - run: python -m bench.run
+
+  package:
+    name: package (${{ matrix.os }})
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [ubuntu-24.04, windows-2025, macos-15]
+    steps:
+      - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
+        with:
+          persist-credentials: false
+      - uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: "3.14"
+      - name: Install build validation tools
+        run: python -m pip install build twine
+      - name: Build wheel and source distribution
+        run: python -m build
+      - name: Validate and smoke-test distributions
+        run: python .github/scripts/check_distributions.py
+
+  # Sift is also used as a data gate against its own examples.
+  data-gate:
+    runs-on: ubuntu-24.04
+    steps:
+      - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
+        with:
+          persist-credentials: false
+      - uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: "3.12"
+      - run: pip install -e ".[all]"
+      - name: Lint the example files
+        run: sift-lint check examples/ --fail-on none
+      - name: Compare against the baseline
+        run: sift-lint diff examples/orders_baseline.csv examples/orders_messy.csv --fail-on none
+      - name: Repair what is safely repairable
+        run: sift-lint fix examples/orders_messy.csv --force --dry-run
+      - name: Report the blast radius
+        run: sift-lint impact examples/orders_messy.csv --sum order_total --group-by region

sift_lint-1.0.0/.github/workflows/release.yml

@@ -0,0 +1,68 @@
+name: release
+
+on:
+  release:
+    types: [published]
+
+permissions:
+  contents: read
+
+jobs:
+  build:
+    if: ${{ !github.event.release.prerelease && !github.event.release.draft }}
+    runs-on: ubuntu-24.04
+    permissions:
+      contents: read
+    steps:
+      - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
+        with:
+          ref: ${{ github.event.release.tag_name }}
+          fetch-depth: 0
+          persist-credentials: false
+      - uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: "3.14"
+      - name: Verify release tag and source
+        env:
+          RELEASE_TAG: ${{ github.event.release.tag_name }}
+        shell: bash
+        run: |
+          PYTHONPATH=. python .github/scripts/check_release.py "$RELEASE_TAG"
+          git fetch --no-tags origin main
+          git merge-base --is-ancestor HEAD origin/main
+          test "$(git rev-parse HEAD)" = "$GITHUB_SHA"
+      - name: Install build and validation tools
+        run: python -m pip install -e ".[dev]" build twine
+      - name: Run release checks
+        run: |
+          ruff check .
+          pytest -q
+          python -m bench.run
+      - name: Build distributions
+        run: python -m build
+      - name: Validate installed distributions
+        run: python .github/scripts/check_distributions.py
+      - name: Upload distributions
+        uses: actions/upload-artifact@cf430e030ddbb5b0abf93d22962f4752f3646cd9 # v7.0.2
+        with:
+          name: release-distributions
+          path: dist/
+          if-no-files-found: error
+
+  publish:
+    if: ${{ !github.event.release.prerelease && !github.event.release.draft }}
+    needs: build
+    runs-on: ubuntu-24.04
+    environment:
+      name: pypi
+      url: https://pypi.org/p/sift-lint
+    permissions:
+      id-token: write
+    steps:
+      - name: Download validated distributions
+        uses: actions/download-artifact@9000827ccba6bdab643e8b6fd33ac0654aef8333 # v8.0.2
+        with:
+          name: release-distributions
+          path: dist/
+      - name: Publish distributions to PyPI
+        uses: pypa/gh-action-pypi-publish@dc37677b2e1c63e2034f94d8a5b11f265b73ba33 # v1.14.2

sift_lint-1.0.0/.gitignore

@@ -0,0 +1,25 @@
+__pycache__/
+*.pyc
+.pytest_cache/
+.ruff_cache/
+.coverage
+.coverage.*
+htmlcov/
+dist/
+build/
+*.egg-info/
+*.whl
+.venv/
+venv/
+.DS_Store
+report.html
+
+# Artefacts the examples in the README produce when you run them
+examples/orders_repaired.csv
+examples/repair-log.csv
+examples/*-report.txt
+examples/learned-sift.toml
+
+# Recording artefacts (the finished GIF is committed; working files are not)
+demo/clean.csv
+demo/*.cast

sift_lint-1.0.0/BENCHMARK.md

@@ -0,0 +1,40 @@
+# Benchmark
+
+Generated by `python -m bench.run`. A clean file is generated, verified
+silent, then corrupted one way at a time; each corruption declares the
+finding code that should catch it.
+
+**Recall: 28/28 (100%)**
+
+**Noise on the clean control file: 0**
+
+| Corruption | Expected finding | Result |
+| --- | --- | --- |
+| -999 placeholders | `numeric-sentinel` | caught |
+| CRLF and LF mixed | `mixed-line-endings` | caught |
+| European decimal separators | `european-numbers` | caught |
+| N/A instead of empty | `disguised-null` | caught |
+| UTF-8 BOM | `bom` | caught |
+| ambiguous date order | `ambiguous-dates` | caught |
+| bad encoding round-trip | `mojibake` | caught |
+| column collapsed to one value | `constant-column` | caught |
+| column emptied entirely | `empty-column` | caught |
+| column with no name | `unnamed-column` | caught |
+| currency formatting | `number-as-text` | caught |
+| dates in the future | `future-dates` | caught |
+| epoch placeholder dates | `sentinel-dates` | caught |
+| extreme outlier | `outliers` | caught |
+| half the column emptied | `missing-values` | caught |
+| headers differing by case | `similar-column-names` | caught |
+| key value repeated | `key-not-unique` | caught |
+| mixed casing | `label-variants` | caught |
+| mixed decimal conventions | `conflicting-number-formats` | caught |
+| repeated rows | `duplicate-rows` | caught |
+| row with an extra field | `ragged-rows` | caught |
+| single misspelling | `near-duplicate-labels` | caught |
+| text in a number column | `mixed-types` | caught |
+| thousands separators | `number-as-text` | caught |
+| trailing spaces | `whitespace` | caught |
+| two columns named the same | `duplicate-column-name` | caught |
+| two date formats mixed | `conflicting-date-formats` | caught |
+| zero-padded codes | `leading-zeros` | caught |

sift_lint-1.0.0/CHANGELOG.md

@@ -0,0 +1,30 @@
+# Changelog
+
+Notable user-visible changes to Sift are recorded here. Releases use
+[Semantic Versioning](https://semver.org/).
+
+## 1.0.0
+
+### Included in the first stable distribution
+
+- `sift-lint` CLI with `check`, `diff`, `fix`, `impact`, `init`, and `profile`
+  commands; a small Python API is available as `sift_lint`.
+- Dependency-free CSV linting, with optional Excel and Parquet readers
+  provided through the `excel`, `parquet`, and `all` extras.
+- Stable finding codes with text, JSON, and standalone HTML reports;
+  cross-file reference checks and configuration through `sift.toml`.
+- Conservative repair with an audit log, baseline drift analysis, and
+  quantified aggregate-impact reporting.
+- Terminal-control escaping and sensitive-data redaction safeguards.
+
+### Release readiness
+
+- Cross-platform CI covers Python 3.11–3.14 on Linux and Python 3.14 on
+  Windows/macOS; wheel and source-distribution smoke tests run on all three.
+- The controlled corruption benchmark detects 28 of 28 planted defects
+  with zero findings on its clean control. This is not a general recall claim.
+- Secure release-only PyPI publishing uses GitHub Actions and short-lived
+  Trusted Publishing credentials; GitHub Actions dependencies use pinned SHAs
+  with Dependabot updates.
+- Public installation instructions, package metadata, contribution guidance,
+  and a security-reporting policy accompany the distribution.

sift_lint-1.0.0/CODES.md

@@ -0,0 +1,127 @@
+# Finding codes
+
+Every code Sift can emit. These strings are the stable interface — they appear in `--format json`, and they are what you name to silence a check:
+
+```toml
+[sift]
+ignore = ["outliers"]
+
+[sift.ignore_columns]
+high-cardinality = ["email"]
+```
+
+Silencing per column is almost always better than silencing globally: a `notes` column is *supposed* to be free text, but the same finding on `customer_id` means an identifier leaked into a category field.
+
+`sift-lint init` writes these exemptions for you, with the reason attached.
+
+A test asserts this file lists exactly the codes the source can emit, so it cannot drift out of date.
+
+## Structure
+
+| Code | Severity | Meaning |
+| --- | --- | --- |
+| `ragged-rows` | error | Rows without the expected number of fields — usually an unescaped delimiter or an unclosed quote. |
+| `no-data-rows` | error | A header and nothing under it. Readers get an empty result rather than an error. |
+| `null-bytes` | error | NUL bytes in a text file — binary content with the wrong extension, or a truncated write. |
+| `encoding` | error | The file is not valid UTF-8. Read as Latin-1 to continue. |
+| `duplicate-column-name` | error | The same column name appears twice; most readers silently keep one. |
+| `bom` | warning | A UTF-8 byte order mark that some readers fold into the first column name. |
+| `mixed-line-endings` | warning | CRLF and LF in the same file, which makes diffs and checksums noisy. |
+| `unnamed-column` | warning | A column with no name in the header. |
+| `padded-column-name` | warning | A header with surrounding whitespace, so lookups by name miss it. |
+| `similar-column-names` | warning | Two headers differing only by case or spacing. |
+| `no-trailing-newline` | info | Concatenating this file with another would join two rows into one. |
+
+## Types and numbers
+
+| Code | Severity | Meaning |
+| --- | --- | --- |
+| `mixed-types` | error | A column that is mostly numbers or dates, with values that are neither. |
+| `conflicting-number-formats` | error | The column mixes English and European decimal conventions; one of them is wrong by a thousand. |
+| `ambiguous-decimal-separator` | error | Values like `1.234` differ by 1000x between conventions and nothing settles which was meant. |
+| `european-numbers` | warning | Comma decimal, dot thousands. An English-locale reader misreads these silently. |
+| `number-as-text` | warning | Numbers written with currency symbols, separators, percent signs or accounting parentheses. |
+| `leading-zeros` | warning | An all-digit column whose zeros a numeric read would destroy. |
+| `numeric-sentinel` | warning | A repeated placeholder like `-999` that will be averaged as a real value. |
+| `outliers` | info | Values far from the median by modified z-score. Silent on skewed columns. |
+| `unexpected-negative` | info | Negative values in a column named like a quantity. |
+| `compact-date-or-number` | info | Every value is 8 digits, equally readable as `YYYYMMDD` or a plain number. |
+
+## Dates
+
+| Code | Severity | Meaning |
+| --- | --- | --- |
+| `conflicting-date-formats` | error | Some rows can only be day-first and others only month-first. Unrepairable. |
+| `ambiguous-dates` | error | Every value reads both ways and nothing in the column settles the order. |
+| `weak-date-evidence` | warning | An order was inferred, but only a small share of the column proves it. |
+| `mixed-date-formats` | warning | ISO dates mixed with slash-separated ones. |
+| `future-dates` | warning | Dates after today. |
+| `implausible-dates` | warning | Dates before 1900. |
+| `sentinel-dates` | warning | Placeholders such as `1970-01-01` or `9999-12-31` treated as real dates by range filters. |
+
+## Missing data
+
+| Code | Severity | Meaning |
+| --- | --- | --- |
+| `missing-values` | error / warning | Null rate above the configured threshold. Error above `null_error`, warning above `null_warn`. |
+| `disguised-null` | warning | Missing values written as text (`N/A`, `unknown`, `-`) so they load as strings. |
+| `empty-column` | warning | A column with no values at all. |
+| `constant-column` | info | One value in every row — check whether a filter upstream collapsed it. |
+
+## Values
+
+| Code | Severity | Meaning |
+| --- | --- | --- |
+| `formula-injection` | warning | Values starting with `=`, `+` or `@`, which a spreadsheet executes as a formula rather than showing as text. |
+| `mojibake` | error | Text mangled by a bad encoding round-trip. Not repairable; re-export the source. |
+| `label-variants` | warning | The same category spelled several ways by case or spacing. |
+| `whitespace` | warning | Stray, doubled or non-breaking whitespace inside values. |
+| `near-duplicate-labels` | info | A rare misspelling of a spelling the rest of the column agrees on. |
+| `high-cardinality` | info | A text column that is almost entirely unique — free text or an identifier, not a category. |
+
+## Keys and references
+
+| Code | Severity | Meaning |
+| --- | --- | --- |
+| `key-not-unique` | error | A declared key repeats, so joining on it multiplies rows. |
+| `key-null` | error | A declared key is empty in some rows. |
+| `missing-key-column` | error | A declared key column is not in the file. |
+| `orphaned-reference` | error | Values pointing at rows that do not exist in the referenced file. |
+| `reference-not-unique` | error | The far side of the relationship repeats, so it cannot identify a row. |
+| `reference-matches-after-cleaning` | warning | Orphans that would match if case and spacing were normalised — dirty keys, not missing records. |
+| `reference-null` | warning | The referencing column is empty, so those rows join to nothing. |
+| `candidate-key` | info | A unique, complete column that could be enforced with `--key`. |
+
+## Sensitive data
+
+Nothing here is a defect — the file is fine. These say *this file contains personal or financial data*, which matters when it is about to be emailed, committed, or fed into a model. **No finding in this group ever includes example values**: a tool that prints someone's card number into a CI log has made the problem worse.
+
+| Code | Severity | Meaning |
+| --- | --- | --- |
+| `sensitive-data` | warning | A column holds email addresses, phone numbers, payment card numbers or IBANs. Cards require an issuer prefix, a valid length and a Luhn check; phone numbers require an international prefix, or a local shape in a column named like a phone field. |
+| `sensitive-column-name` | info | A column is *named* like a sensitive field (`ssn`, `date_of_birth`, `iban`). The values were not inspected. |
+
+## Excel and Parquet
+
+| Code | Severity | Meaning |
+| --- | --- | --- |
+| `formula-error` | error | `#REF!`, `#N/A` and similar exported as literal text. |
+| `partial-excel-dates` | error | Some cells are real dates and the rest are text that looks identical on screen. |
+| `preamble-rows` | warning | Title or spacer rows above the header, written for a human reader. |
+| `loose-schema` | warning | A Parquet column declared as string whose values are all numbers. |
+| `unread-sheets` | info | Other worksheets in the workbook that nothing checked. |
+
+## Drift, from `sift-lint diff`
+
+| Code | Severity | Meaning |
+| --- | --- | --- |
+| `column-removed` | error | A baseline column is gone. |
+| `type-changed` | error | A column's inferred type changed between the two files. |
+| `new-category` | warning | Categories not present in the baseline. |
+| `null-rate-jump` | warning | A column became substantially emptier. |
+| `distribution-shift` | warning | The median moved enough to suggest a unit change. |
+| `column-order-changed` | warning | Harmless for named access, fatal for anything reading by position. |
+| `row-count-shift` | warning | The file grew or shrank dramatically. |
+| `column-added` | info | A column not present in the baseline. |
+| `missing-category` | info | Baseline categories that no longer appear. |
+| `duplicate-rows` | warning | Exact repeated rows, which inflate any total. |

sift_lint-1.0.0/CONTRIBUTING.md

@@ -0,0 +1,46 @@
+# Contributing to Sift
+
+Bug reports, reproducible false positives, documentation corrections and
+focused pull requests are welcome. For security vulnerabilities, follow
+[SECURITY.md](SECURITY.md) rather than filing a public issue.
+
+## Before proposing a change
+
+Check existing [issues](https://github.com/N0tDeb/sift/issues). A useful data
+quality bug report includes a small, non-sensitive input file, the command
+used, the expected result and the actual finding or error. Do not attach
+real customer data or secrets.
+
+Sift's priority is accurate findings with few false positives. Before adding
+a rule, explain what downstream damage it detects and when correct input
+could produce a false alarm. Repairs must not guess when evidence is unclear.
+
+## Development
+
+Sift needs Python 3.11 or newer. After cloning, install development extras
+in a suitable environment:
+
+```bash
+python -m pip install -e '.[dev]'
+ruff check .
+pytest -q
+python -m bench.run
+```
+
+The `dev` extra supplies pytest, Ruff and both optional file readers. No
+special local release tooling is required; distribution checks run in GitHub
+Actions. See [SETUP.md](SETUP.md) for the CI and maintainer workflow.
+
+## Pull requests
+
+- Branch from the latest `main` and keep changes focused.
+- Add a regression test for behavioral fixes, including cases where Sift
+  should produce **no** finding.
+- For new checks, extend the known-corruption benchmark where applicable and
+  update [CODES.md](CODES.md) when finding codes change.
+- Update README examples and [Project Guide](docs/PROJECT_GUIDE.md) when
+  public commands, supported formats or expectations change.
+- Make sure CI passes on Windows, macOS and Linux before requesting a merge.
+
+The public command is `sift-lint`; the Python package is `sift_lint`.
+See the [MIT license](LICENSE) for contribution licensing.

sift_lint-1.0.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Sift contributors
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.