NeuNorm 2.2.0__tar.gz → 2.2.2__tar.gz

{neunorm-2.2.0 → neunorm-2.2.2}/.grype.yaml

@@ -28,3 +28,7 @@ ignore:
   - vulnerability: CVE-2026-7774
     package:
       name: python
+  # medium; fixed in the 3.13 line + 3.15 beta, not the 3.14 line the build uses
+  - vulnerability: CVE-2026-12003
+    package:
+      name: python

{neunorm-2.2.0 → neunorm-2.2.2}/CHANGELOG.md

@@ -5,7 +5,48 @@ All notable changes to NeuNorm are documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
-## [Unreleased]
+## [2.2.2] - 2026-07-02
+
+### Fixed
+
+- **`background_roi` can now reproduce legacy 1.x zero-count-ROI semantics** (completes the
+  downstream-superset goal of
+  [#172](https://github.com/ornlneutronimaging/NeuNorm/issues/172) /
+  [#159](https://github.com/ornlneutronimaging/NeuNorm/issues/159)): the pooled-mean
+  strictly-positive/finite guard can be opted out with
+  `apply_background_roi(..., strict=False)` or
+  `normalize_transmission(...)` / `normalize_with_dark(..., background_roi_strict=False)`,
+  letting a zero-count background ROI propagate `inf`/`nan` through the division exactly as
+  NeuNorm 1.x did (iBeatles pins this behavior). The default stays strict — a
+  non-positive/non-finite pooled mean raises `ValueError` — and structural errors (bad ROI
+  bounds, missing dims) always raise.
+
+## [2.2.1] - 2026-07-01
+
+### Added
+
+- **`background_roi` now supports pooled multiple ROIs, inclusive extents, and an open-beam-less
+  mode.** `normalize_transmission(..., background_roi=)` — and the MARS CCD/TPX3 and VENUS CCD
+  pipelines — accept a **sequence** of ROIs, pooled as `sum(counts) / sum(pixels)` per image;
+  `ROI(..., inclusive=True)` opts into legacy inclusive extents (a width-`w` ROI spans `w+1`
+  pixels); and the new `neunorm.processing.normalizer.apply_background_roi(data, background_roi)`
+  applies the flux proxy to a sample stack with no open beam. A single ROI / bare tuple is
+  unchanged. This makes `background_roi` a superset of the downstream (iBeatles) pooled-inclusive
+  re-implementation so it can be removed.
+  ([#172](https://github.com/ornlneutronimaging/NeuNorm/issues/172),
+  [#159](https://github.com/ornlneutronimaging/NeuNorm/issues/159))
+- **Python 3.14 support.** NeuNorm builds, installs, and passes its full test suite on
+  CPython 3.14; the development/CI pixi environments and `pixi.lock` move to 3.14
+  (`requires-python` stays `>=3.11`).
+
+### Fixed
+
+- **TIFF export is compatible with scitiff ≥ 26.6.** scitiff 26.6 tightened its metadata schema
+  to reject object-dtype (`PyObject`) scipp variables. `write_tiff_stack` now JSON-encodes
+  sequence metadata (mirroring the HDF5 writer's provenance convention — read it back with
+  `json.loads`) and drops object-dtype coords/masks (e.g. tuple-valued TIFF header tags carried
+  over from the input files) that scitiff cannot serialize. HDF5 output and the written image
+  data are unchanged.
 
 ## [2.2.0] - 2026-06-30
 
@@ -171,6 +212,8 @@ documentation are archived under
 [`archive/neunorm-1.x/`](archive/neunorm-1.x/); released 1.x versions remain
 available on PyPI and the `conda-forge` channel (`pip install "NeuNorm<2"`).
 
+[2.2.2]: https://github.com/ornlneutronimaging/NeuNorm/releases/tag/v2.2.2
+[2.2.1]: https://github.com/ornlneutronimaging/NeuNorm/releases/tag/v2.2.1
 [2.2.0]: https://github.com/ornlneutronimaging/NeuNorm/releases/tag/v2.2.0
 [2.1.0]: https://github.com/ornlneutronimaging/NeuNorm/releases/tag/v2.1.0
 [2.0.0]: https://github.com/ornlneutronimaging/NeuNorm/releases/tag/v2.0.0

{neunorm-2.2.0 → neunorm-2.2.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: NeuNorm
-Version: 2.2.0
+Version: 2.2.2
 Summary: Neutron imaging normalization and TOF data processing for ORNL facilities
 Project-URL: homepage, https://github.com/ornlneutronimaging/NeuNorm
 Project-URL: repository, https://github.com/ornlneutronimaging/NeuNorm

{neunorm-2.2.0 → neunorm-2.2.2}/docs/migration.md

@@ -106,7 +106,7 @@ transmission = normalize_transmission(sample, ob)  # T = sample / ob, variances
 | `Normalization()` (stateful object) | a `run_*_pipeline(...)` call, or composable functions |
 | `.load(..., data_type="sample"/"ob")` | pipeline `sample_paths` / `ob_paths`, or `neunorm.loaders.stack_loader.load_stack` / `tiff_loader.load_tiff_stack` / `fits_loader.load_fits_stack` |
 | `.load(..., data_type="df")` (dark/"df") | pipeline `dark_paths`, or `neunorm.processing.dark_corrector.subtract_dark(data, dark)` |
-| `.normalization(roi=...)` | Flux normalization by a background ROI: `normalize_transmission(..., background_roi=(x0, y0, x1, y1))`, or pass `background_roi=` to `run_mars_ccd_pipeline` / `run_mars_tpx3_pipeline` / `run_venus_ccd_pipeline` ([#159](https://github.com/ornlneutronimaging/NeuNorm/issues/159)). A proton-charge proxy for when proton charge is unavailable; mutually exclusive with it. Note: 2.x `roi=` is a spatial **crop** (`apply_roi`), not this. |
+| `.normalization(roi=...)` | Flux normalization by a background ROI: `normalize_transmission(..., background_roi=(x0, y0, x1, y1))`, or pass `background_roi=` to `run_mars_ccd_pipeline` / `run_mars_tpx3_pipeline` / `run_venus_ccd_pipeline` ([#159](https://github.com/ornlneutronimaging/NeuNorm/issues/159)). A proton-charge proxy for when proton charge is unavailable; mutually exclusive with it. Accepts one ROI or a **pooled sequence** of ROIs; use `ROI(..., inclusive=True)` for the 1.x inclusive extents, and `apply_background_roi(data, background_roi)` for the open-beam-less form. Note: 2.x `roi=` is a spatial **crop** (`apply_roi`), not this. |
 | `.df_correction()` | `neunorm.processing.dark_corrector.subtract_dark(data, dark)` |
 | `.crop(roi=ROI(...))` | `neunorm.processing.roi_clipper.apply_roi(data, ROI(x0=.., y0=.., x1=.., y1=..))` (or a bare `(x0, y0, x1, y1)` tuple) |
 | auto/manual gamma filtering on `load()` | `neunorm.filters.gamma_filter.apply_gamma_filter(...)`, or pipeline `gamma_filter=True` |