[Paullo9]

Requested by Paul · project thread

Part of #10 (item 2 of the 29 Oct plan). Independent of #43 (item 1). Still open from #10:

• item 3: partial PixelSize/Offset, detector name and Ceta date, re-measured against RosettaSciIO on the real files first;
• item 4: camera pixel pitch. Found while looking: none of RosettaSciIO's 15 public Velox files states a physical camera pitch (Detectors.* carries only Binning), and RosettaSciIO maps DM's Acquisition.Device.CCD.Pixel Size (um) for CL signals only, so item 4 needs a file that states it before a field is added.

What changes

• VeloxSpectrumSignals maps Sample.elements as HyperSpy sets it when it applies RosettaSciIO's _convert_element_list (d.keys()[0] on HyperSpy's DictionaryTreeBrowser: the first of the sorted, slugified keys; atomic_number2name 1 to 119; SharedProperties/EDSSpectrumQuantificationSettings from version 11, Operations/ImageQuantificationOperation before).
• The four file-level groups (Displays, Operations, SharedProperties, Features) are parsed as read_file parses them (shared with VeloxTitles); the read is refused wherever RosettaSciIO or HyperSpy raises, int(version) and the version key included.
• FormatoValue.texts([String]) holds the list (with a texts accessor).
• VeloxEMDReader.spectrumMetadata() and VeloxMetadata.binaryResultSegmentTimes: each spectrum's live and real time as Velox 4 to 13 store them per segment in Detectors (RosettaSciIO maps only DetectorMetadata.*, which no public spectrum holds). Not in the signals' metadata, which stays RosettaSciIO's.
• FormatoLoader.load keeps a Velox file's images when its spectra are refused (as it already does for spectrum images).
• docs/parity.md and docs/open-items.md updated, with the deviations left.

Migration

An exhaustive switch over FormatoValue needs a .texts arm (or default). The suite's apps have none.

Tests

• tools/velox-elements/pins.py: 20 planted files, read by RosettaSciIO 049e7d70 with the mapping applied by HyperSpy 2.4.0's own DictionaryTreeBrowser, plus the public files (Test SI 16x16 215 kx.emd gives ["C", "O", "Ca", "Cu"], as RosettaSciIO's own test).
• VeloxSpectrumElementsTests: the elements exact or the refusal (unsupported, "RosettaSciIO raises") for every file, a case whose slugified order differs from the raw order, a slugify table checked against HyperSpy, the segment times to the bit, and the public files when unzipped.
• swift test -c release --parallel --skip FormatoKitParityTests: green, except the HDF5 exit-barrier test, which needs FORMATO_HDF5_PATH from tools/test.sh (environment, not this change).
• Reviewed before commit (public API). The findings are fixed: slug-order fixture, specific refusal checks, _-prefixed keys, duplicate slugs refused, HDF5 errors no longer hidden in spectrumMetadata(), docs.

🤖 Generated with Claude Code