cranalytics 0.2.0__tar.gz

cranalytics-0.2.0/.gitignore

@@ -0,0 +1,182 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+.python-version
+
+# pipenv
+Pipfile.lock
+
+# poetry
+poetry.lock
+
+# pdm
+.pdm.toml
+.pdm-python
+
+# PEP 582
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# VS Code
+.vscode/
+
+# JetBrains IDEs
+.idea/
+
+# macOS
+.DS_Store
+
+# Windows
+Thumbs.db
+ehthumbs.db
+*.db
+
+# Local review notes
+/doc.md
+RECOMMENDATIONS_ACTION_PLAN.md
+
+# Generated pipeline reports
+*_pipeline_report.html
+*.png
+
+# Git worktrees
+.worktrees/
+
+# Locally installed agent skills (managed by npx skills, not committed)
+.agents/
+.agent/
+
+# Claude Code local files (not project config)
+.claude/settings.local.json
+.claude/skills/
+CLAUDE.local.md
+
+# Codex worktree artifacts
+.codex-worktrees/
+
+# Temporary test output — all temp dirs go here, not project root
+.tmp/
+tests_tmp/
+.pytest_tmp/
+manual_check_tmp/
+pytest_tmp*/
+pytest-cache-files-*/

cranalytics-0.2.0/CHANGELOG.md

@@ -0,0 +1,300 @@
+# Changelog
+
+All notable changes to this project will be documented in this file.
+
+The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
+and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [Unreleased]
+
+## [0.2.0] - 2026-06-10
+
+### Removed (breaking)
+
+- **31 deprecated top-level exports.** `from cranalytics import <name>` now raises an
+  `AttributeError` tombstone naming the workflow module to import from instead.
+  Migration table:
+
+  | Old import (`from cranalytics import ...`) | New import |
+  |---|---|
+  | `calculate_early_performance_rates`, `compute_conditional_loss_table`, `compute_marginal_impact`, `compute_segment_rates`, `estimate_vintage_lifetime_profit`, `rank_features_by_separation`, `validate_performance_flags` | `from cranalytics.early_performance import ...` |
+  | `FPFWorkflowReport`, `print_fpf_report`, `run_fpf_workflow`, `train_fpf_challenger` | `from cranalytics.fpf_workflow import ...` |
+  | `ModelSweepSpec`, `build_curve_fitter_sweep_spec`, `run_backtest_sweeps`, `select_champion_and_challengers`, `summarize_variant_performance` | `from cranalytics.method_selection import ...` |
+  | `run_rollforward_backtest_sweeps`, `select_rollforward_champion_and_challengers`, `summarize_rollforward_variant_performance` | `from cranalytics.rollforward_backtest import ...` |
+  | `calibrate_score_to_event_rate`, `compute_actual_vs_expected`, `compute_psi`, `score_performance_monitoring_report`, `simulate_policy_cutoff` | `from cranalytics.score_monitoring import ...` |
+  | `compare_known_actuals_to_curves`, `fit_flow_hazard_curves`, `forecast_balance_flows`, `validate_flow_data` | `from cranalytics.survival_flows import ...` |
+  | `compute_cgco_curve_wide`, `compute_final_cgco_wide`, `load_wide_vintage_data` | `from cranalytics.vintage_wide import ...` |
+
+- **`rollforward-workflow` and `rollforward-readiness` CLI commands.** Use
+  `cranalytics rollforward` and `cranalytics rollforward --mode readiness`; the old
+  spellings now exit with a parser error.
+
+### Deprecated
+
+- **`cranalytics.transition`** is now a shim emitting `DeprecationWarning`; it will be
+  removed in 0.3.0. The estimator implementation moved to the private
+  `cranalytics._transition_estimation` module. Supply a transition matrix, transition
+  ledger, or loan history directly to `forecast_lifetime_loss` instead of estimating
+  cohort matrices yourself.
+
+### Changed
+
+- The top-level namespace is now 51 names (22 preferred + 29 compatibility), pinned by
+  `tests/test_public_api_surface.py`.
+- `fpf_workflow` is documented as an applied workflow rather than core API; its
+  functions are importable only from `cranalytics.fpf_workflow`.
+
+## [0.1.22] - 2026-06-07
+
+### Added
+- Shared rollforward result-summary types in `src/cranalytics/rollforward_results.py` so `run_rollforward(...)`, `run_rollforward_workflow(...)`, and `generate_rollforward_readiness_report(...)` all expose the same `run_summary` / `operational_status` vocabulary.
+- Canonical rollforward output payload adapters in `src/cranalytics/_rollforward_output_payloads.py` plus dedicated readiness/workflow writer modules for surface-specific artifact emission.
+- New rollforward artifacts `rollforward_run_summary.json` and `workflow_run_metadata.json` for readiness and workflow outputs, alongside governance-model adapters that consume the shared run-summary contract.
+- New characterization and regression coverage for shared rollforward status semantics, result adapters, reporting payloads, and workflow/result typing.
+
+### Changed
+- Consolidated the rollforward public surfaces around typed internal session results in `src/cranalytics/_workflow_results.py` and shared public result adapters in `src/cranalytics/rollforward_results.py`, leaving `rollforward.py`, `rollforward_workflow.py`, and `rollforward_readiness.py` as thinner orchestration wrappers.
+- Centralized rollforward session-to-public-result translation and JSON-ready payload shaping, then split artifact writing by surface into `src/cranalytics/_rollforward_readiness_reporting.py` and `src/cranalytics/_rollforward_workflow_reporting.py`.
+- Surfaced shared rollforward operational status in CLI output, committee summaries, champion-selection payloads, readiness summaries, the core rollforward example, and the output-contract documentation.
+- Reframed FPF as an advanced binary actual-performance-flag workflow in docs, examples, and onboarding paths instead of treating it as a front-door package workflow.
+- Continued the top-level API curation pass by moving the remaining docs and bundled-skill examples toward module-local imports and preserving the root namespace mainly as a compatibility surface.
+
+### Deprecated
+- Additional top-level convenience imports for advanced helpers now emit `DeprecationWarning`, including FPF-specific workflow symbols, early-performance helpers, and survival-flow utilities. Callers should import these from their workflow modules directly.
+
+### Removed
+- Removed the internal `src/cranalytics/rollforward_reporting.py` facade after all tracked callers and tests moved to the shared payload layer plus the dedicated readiness/workflow reporting modules.
+
+### Fixed
+- Unified rollforward `ok` / `data_issues` / `insufficient_data` semantics across in-memory results, readiness outputs, workflow outputs, and written artifacts so the same session state no longer needs to be re-derived in multiple places.
+- Preserved `max_mob`, issue counts, and operational-status details consistently across rollforward facade, workflow, and readiness summaries.
+
+## [0.1.21] - 2026-06-05
+
+### Added
+- Workflow-aware `cranalytics validate-data` support with auto-detection and dispatch for vintage, feature-analytics, and related CSV shapes via `validation_dispatch.py`.
+- Scriptable `cranalytics quickstart` flags: `--workflow`, `--show-requirements`, and `--write-template`, plus starter CSV templates for vintage and feature-analytics workflows.
+- Read-only `cranalytics install-skills --list` and `--describe` modes so bundled skills can be discovered without performing an install.
+- Shared internal loan-tape normalization helpers in `src/cranalytics/_loan_tape_normalization.py` so loan snapshot and loan history contracts stop cross-importing private helpers.
+
+### Changed
+- Centralized workflow onboarding metadata and generated workflow-routing docs from a single registry, simplifying `quickstart`, CLI help, and starter-template wiring.
+- Narrowed the documented top-level API promise: `src/cranalytics/__init__.py` now distinguishes a small preferred convenience surface from the wider compatibility namespace.
+- Top-level imports for specialized method-selection, score-monitoring, rollforward-backtest, and wide-vintage helpers now emit `DeprecationWarning` and point callers to workflow-module imports.
+- Tightened rollforward workflow session typing and explicit status=`"ok"` artifact guards so downstream reporting and export paths no longer rely on optional outputs implicitly being present.
+- Updated the repo-local `publish-package` skill to use the dynamic version source in `src/cranalytics/__init__.py` and the correct `cranalytics` workspace/import names.
+
+### Fixed
+- Hardened workflow-aware validation contracts so feature-analytics CSV validation rejects invalid balances, rates, terms, dates, duplicate identifiers, and unsupported target values while preserving blank immature targets.
+- Aligned vintage `validate-data` auto-detection and validation with the documented minimal unsegmented shape used by starter templates and onboarding flows.
+- Cleaned tracked Ruff issues and reduced source `basedpyright` output to a single known `__all__` warning in `src/cranalytics/__init__.py`.
+- Resolved pandas typing and guard issues across loan contract normalization, score monitoring, forecasting bridge, validation, survival flows, and vintage transforms uncovered during the release-hardening pass.
+
+## [0.1.20] - 2026-04-10
+
+### Added
+- Ledger-first Loss Forecasting support: `forecast_lifetime_loss(...)` and `summarize_lifetime_loss(...)` now accept transition ledgers and loan-history panels in addition to explicit square matrices.
+- Governance metadata models in `src/cranalytics/governance_models.py` for non-DataFrame workflow metadata validation, backed by `pydantic>=2`.
+- Mapping-compatible `PredictiveModelingSessionResult` and `VintageAnalysisSessionResult` outputs with expanded docs and examples around `run_predictive_modeling_session(...)` and `run_vintage_analysis_session(...)`.
+
+### Changed
+- Deepened predictive and vintage session boundaries so workflow callers receive clearer typed session seams instead of looser dict-style orchestration paths.
+- Refreshed the Lifetime Loss Forecasting, ML modeling, and vintage tutorials/examples to align with the ledger-first and typed-session workflow boundaries.
+
+## [0.1.19] - 2026-04-08
+
+### Added
+- `project_incomplete_vintage_tails(df, incomplete_vintages, smoother, max_mob)` — projects forward-looking loss curves for immature vintages using a fitted smoother; clips projections to non-decreasing and ≤ 1.0. Exported as `cranalytics.project_incomplete_vintage_tails`.
+- `run_vintage_analysis_session` gains `extrapolate_tails` and `tail_max_mob` parameters; result dict always includes `"tail_projections"` key.
+
+### Changed
+- `score_model` argument order changed to DataFrame-first: `score_model(df, estimator, feature_cols, ...)`. Old order `(estimator, df, ...)` still works with a `DeprecationWarning`. Enables clean `.pipe()` chaining.
+- Internal DataFrame transforms in `calculate_duration_and_event` (`survival.py`) and `evaluate_rollforward_validation_window` (`rollforward_evaluation.py`) converted from in-place mutation to `.assign()` chains.
+
+## [0.1.18] - 2026-04-05
+
+### Added
+- Internal `LoanState` enum coverage across datasets, simulation, and Loss Forecasting to make not-yet-due and terminal-state handling more explicit.
+- `CurveFitter.converged` tracking and backtest-schema support so model-selection workflows can distinguish failed fits from valid challengers.
+- New regression suites for curve-fitter convergence, forecasting-bridge conservation, forecasting-bridge vectorization, loan states, and survival-flow guards/vectorization.
+
+### Changed
+- Vectorized `forecast_balance_flows(...)` and `forecast_calendar_chargeoff_from_predictions(...)` to remove row-wise iteration from the main allocation/forecasting paths.
+- Consolidated shared helper logic such as `_require_columns`, `_validate_confidence`, and `_months_elapsed` under `src/cranalytics/_helpers.py`.
+
+### Fixed
+- Added allocation-conservation warnings so calendar charge-off forecasts surface silent dollar-allocation drift.
+- Guarded fitted hazard curves by clipping rates to `[0, 1]` and raising on NaN/Inf outputs before returning them to callers.
+- Hardened model-selection comparison logic so champion/challenger selection filters infinite-MAPE rows and raises when every variant fails.
+
+## [0.1.17] - 2026-03-19
+
+### Added
+- Package-level module execution entrypoint via `python -m cranalytics`, matching the existing console-script CLI behavior.
+
+## [0.1.16] - 2026-03-19
+
+### Added
+- Onboarding workflow metadata and regression coverage so the quickstart and onboarding path stay aligned as docs and CLI copy evolve.
+- New reference maps and contributor-navigation docs for the public API, workflows, modules, and onboarding paths.
+
+### Changed
+- Made `cranalytics quickstart` the canonical getting-started path, refreshed onboarding copy, and aligned schema guidance across the CLI, docs, and examples.
+- Added packaged feature-analytics and ML examples/tutorials and published the release as `0.1.16` on TestPyPI.
+
+## [0.1.15] - 2026-03-17
+
+### Added
+- Internal rollforward session boundaries to make readiness, workflow, and artifact-generation orchestration more explicit.
+
+### Changed
+- Documented the repo's multi-agent worktree workflow and operating procedure for contributors.
+
+### Fixed
+- Preserved rollforward-contract denominators across backtests so readiness and workflow comparisons stay on the same population basis.
+- Synchronized `CHANGELOG.md` with the shipped `0.1.14` feature set and republished as `0.1.15` after the prior TestPyPI artifacts had already been uploaded.
+
+## [0.1.14] - 2026-03-17
+
+### Added
+- Predictive modeling toolkit expansion: `engineer_loan_features`, `lift_gain_table`, `fit_woe_binning`, `build_targets`, `assemble_modeling_frame`, `score_model`, `train_binary_model`, `train_regression_model`, predictive backtesting helpers, and calendar charge-off forecasting.
+- Flow Performance Framework workflow support: `make_mock_fpf_data`, `run_fpf_workflow`, `train_fpf_challenger`, `FPFWorkflowReport`, `print_fpf_report`, and a packaged FPF demo.
+- Canonical Rollforward workflow modules for CLI execution, rollforward backtesting, rollforward input contract validation, shared rollforward evaluation, reporting adapters, characterization tests, and internal workflow/readiness session boundaries.
+- Explicit predictive and vintage session boundaries with dedicated contract modules and validation coverage for DataFrame-first orchestration paths.
+- New tutorials and reference docs for core workflows, validation ownership, and TestPyPI installation.
+
+### Changed
+- Modernized packaging to use Hatch dynamic versioning and explicit wheel artifacts so bundled skills and `py.typed` are included consistently in distributions.
+- Restored Python 3.10 support in build and CI coverage, and added optional extras for `ml`, `xgboost`, and `lightgbm` workflows.
+- Tightened temporary artifact handling under `.tmp` and refreshed docs/navigation for the expanded workflow surface area.
+
+### Fixed
+- Preserved rollforward-contract denominators across backtests and aligned readiness/workflow paths on the canonical rollforward input boundary.
+- Hardened readiness and interpretation code paths, including FPF challenger calibration handling and additional rollforward schema/validation checks.
+- Cleared multiple CI linting, formatting, type-checking, and workflow integration issues uncovered during the release hardening pass.
+
+## [0.1.13] - 2026-03-13
+
+### Added
+- Early performance analytics module (`early_performance.py`) with flag validation, portfolio/segment rate analysis, WoE/IV, feature ranking, marginal impact, and conditional loss tables.
+- Score monitoring module (`score_monitoring.py`) with score-to-event calibration, actual-vs-expected monitoring, and PSI drift analysis.
+- Visualization module (`viz_early_performance.py`) with eight early-performance chart helpers.
+- Validation factory `make_performance_flag_schema` for dynamic binary flag validation via Pandera.
+- Decision-support helpers including `simulate_policy_cutoff`, `score_performance_monitoring_report`, and `estimate_vintage_lifetime_profit`.
+- New tests for early performance and score monitoring workflows.
+
+## [0.1.12] - 2026-03-12
+
+### Added
+- `cranalytics quickstart` CLI command: 7-item interactive menu with narrated workflow demos and data requirement tables.
+- `src/cranalytics/quickstart.py`: menu logic, six workflow walkthroughs, and CSV template writer.
+- Repo-local `publish-package` release skill plus a README example for the LendingClub full-pipeline workflow.
+
+## [0.1.11] - 2026-03-12
+
+### Fixed
+- Simulation `ValueError: assignment destination is read-only` — added `.copy()` on balances array from `to_numpy()` in `simulation.py`.
+- Synced `__init__.py` `__version__` to match `pyproject.toml` (was behind after namespace rename).
+- Updated `.claude/settings.json` lint hook path after `credit_analytics` → `cranalytics` folder rename.
+
+## [0.1.10] - 2026-03-12
+
+### Added
+- Bundled Claude skills (M1–M4): `vintage-loss-curves`, `loss-forecasting`, `portfolio-diagnostics` under `src/cranalytics/skills/` with `SKILL.md` and `references/`.
+- Skills CI workflow (`.github/workflows/skills-ci.yml`): frontmatter lint and `__all__` alignment checks.
+- `.claude/skills/publish-package/SKILL.md` for guided PyPI publish workflow.
+- `uv` install instructions added to README alongside pip.
+
+### Changed
+- Renamed package namespace `credit_analytics` → `cranalytics` across all source, tests, docs, and CI.
+- Bumped version to `0.1.10`.
+
+## [0.1.9] - 2026-03-11
+
+### Added
+- Bootstrap confidence intervals for vintage curve forecasts via `bootstrap_smoother_intervals()`.
+- CRO package review artifact: `CRO_TECHNICAL_REVIEW_RECOMMENDATIONS.md` with prioritized model-risk findings and a 60-day hardening plan.
+
+### Changed
+- Hardened simulation state propagation, rolling-origin validation defaults, dataset reproducibility, survival strictness, and the rollforward-readiness CLI before the `0.1.9` release.
+
+## [0.1.8] - 2026-03-10
+
+### Added
+- Bundled Claude skills workflow: `install-skills` CLI subcommand installs packaged skills to local skills directories.
+- New analytics modules: `method_selection.py` (variant sweep backtesting, champion/challenger), `survival_flows.py` (rollforward-based hazard modeling), `vintage_wide.py` (wide-format vintage loader + CGCO forecaster).
+- Rollforward-based readiness report: `rollforward_readiness.py` with `generate_rollforward_readiness_report()`.
+- `rollforward-readiness` CLI subcommand exposing `generate_rollforward_readiness_report` (fixes P0 contract gap).
+- Competing-risks survival demos and Aalen-Johansen support.
+- `fetch_lending_club()` dataset loader with pooch caching; hash pinned for reproducibility.
+- `resolve_columns()` dynamic column alias registry in `datasets.py`.
+- Schema validation integrated into `simulate_portfolio_cashflows`.
+- `strict` parameter on `calculate_duration_and_event`: negative-duration records now warn+drop by default; `strict=True` raises.
+
+### Changed
+- Replaced `black` formatter with `ruff format` across all tooling and CI.
+- Relaxed mypy config to reflect actual maturity (`allow_untyped_defs`, `ignore_missing_imports`).
+- Pinned external dataset hash in `datasets.py` for reproducible benchmarks.
+- CI now runs `ruff check` + `ruff format --check` + `pytest` on Python 3.10/3.11/3.12.
+
+### Fixed
+- `fetch_lending_club()` output conforms to `PortfolioSchema` canonical status values.
+- Smoothing validation flow and column validation in `vintage_transforms.py`.
+- Snake_case key consistency in test fixtures.
+
+## [0.1.7] - 2026-03-01
+
+### Added
+- Linting, typing, and docstring cleanup across all source modules.
+- `normalize_return_schema` for `generate_survival_data`.
+- Extra column allowance in `PortfolioSchema` validation.
+
+### Changed
+- Improved CLI discoverability for common onboarding tasks and normalized Lending Club status handling/tests around the validation surface.
+
+## [0.1.6] - 2026-02-25
+
+### Added
+- Core examples refactored as standalone scripts: `core_vintage.py`, `core_lifetime_loss.py`, `core_segmentation.py`.
+- Competing-risks survival analysis utilities.
+- Dynamic column alias registry (`COLUMN_ALIASES`, `resolve_columns()`).
+- `fetch_lending_club()` data-loading path for reproducible Lending Club examples and survival testing.
+
+### Changed
+- Simplified README around 3 core workflows and optional advanced modules.
+- Added docs onboarding page: `docs/getting_started.md`.
+- Split optional dependencies into extras: `viz`, `survival`, `full`.
+- Removed `sys.path` mutations from examples.
+- Naming conventions enforced: `residual_analysis` → `analyze_residuals`, `summary_table` → `create_summary_table`, `temporal_validation` → `validate_temporally`.
+- Updated smoothing demo output paths from `/tmp` to local `output/`.
+
+## [0.1.5] - 2026-02-20
+
+### Changed
+- Applied review-driven fixes and status updates across finance, survival, vintage transforms, vintage validation, and visualization docs/tests after the `0.1.4` packaging rename pass.
+- Published a follow-up TestPyPI release as `0.1.5`.
+
+## [0.1.4] - 2026-02-19
+
+### Added
+- Packaged runnable demos for the core vintage, lifetime-loss, and segmentation workflows.
+
+### Changed
+- Normalized user-facing distribution and CLI naming to `cranalytics` for TestPyPI while documenting the separate `credit_analytics` import namespace.
+
+## [0.1.3] - 2026-02-19
+
+### Changed
+- Restored the published package name to `cranalytics` for TestPyPI compatibility while keeping the existing import namespace unchanged.
+- Shipped as a packaging-only follow-up release with no intended library-behavior changes relative to `0.1.2`.
+
+## [0.1.2] - 2026-02-19
+
+### Changed
+- Packaging-only follow-up TestPyPI release with no intended library-code changes relative to `0.1.1`.
+
+## [0.1.1] - 2026-02-19
+
+### Added
+- Initial public package release with DataFrame-first vintage analysis, smoothing, visualization, lifetime-loss forecasting, survival modeling, and financial-metrics utilities.
+- Pandera-backed validation across the public workflow surface, plus early tutorials, demos, and publishing setup for TestPyPI distribution.

cranalytics-0.2.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2024 Your Name
+
+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.