Skip to content

Shimadzu UVProbe .spc

Shimadzu’s UVProbe software (UV-1800, UV-2600, UV-3600 and other UV-Vis and UV-Vis-NIR spectrophotometers) saves a spectrum as a .spc file. It is not a Galactic SPC file, which uses the same extension: it is a compound file (OLE2) with the spectrum as two float64 arrays. OpenReadout returns each data set as a trace, with the software, instrument model, serial number and creation date.

Derived from public files of two depositors and checked against the UVProbe text exports the depositors saved beside them. See docs/provenance/shimadzu-uvprobe.md.

Crate: openreadout-spectro, format id shimadzu-uvprobe (UVPROBE_FORMAT_ID), extension spc, reader UvProbeReader; the dataset is the shared SpectroDataset.

Detection

A compound file (D0 CF 11 E0 A1 B1 1A E1) holding DataStorage1/DataSetGroup/DataSet1/DataSpectrumStorage/Data/X Data.1 → definite. A compound file with the .spc extension and no such stream → likely (opening it then fails as unsupported). Galactic SPC files have no compound-file header and stay with galactic-spc.

Layout

Streams under DataStorage1/DataSetGroup/, one storage DataSet<n> per data set (DataSetGroupHeaderInfo: u32 count; 1 in every development file):

stream content our reading
DataSpectrumStorage/Data/Data Header.1 u32 point count, u32 value code points; value_code: 2 absorbance, 4 reflectance (%); other codes unnamed (unknown_value_code finding)
DataSpectrumStorage/Data/X Data.1, Y Data.1 float64 × points wavelength (nm, from high to low) and values
DataSpectrumStorage/DataHeader/Header Info three float64: first x, last x, step the axis is regular when the x array follows them
DataSetHeaderInfo one-byte-length texts and numbers: data set name (RawData), u32, float64 OLE Automation date (local time), 6 bytes, organization, 8 bytes, software, version, model, series, u32, a second model text, serial number trace name, instrument.software/software_version/model/serial, acquisition.started_at (the data set’s creation, local time)
MethodStorage/PageTexts<n> u32, u32, a title, u32 count, then label and value texts vendor tree (method_pages), not named
DataHistoryStorage/DataSetHistory the creation record listed
Version/File Format Version u32 (0x120) format_version (file format 0x120)

Texts are UTF-8 when valid, else Latin-1; files written on Chinese-language PCs store GBK, which then reads as Latin-1.

What the reader returns

  • One trace per data set: x wavelength in nm, one channel named by the value code (absorbance; reflectance in %; y when unknown); extra: value_code, data_set, data_set_name, data_type UV/VIS SPECTRUM.
  • Experiment: instrument vendor Shimadzu, model, serial, software and version; acquisition start (the data set’s creation time, local, no zone).
  • check: array lengths against the point count, non-finite values; a data set with .2 arrays (several spectra) reads the first and reports unsupported_data_set.

Validation

Six development files with UVProbe’s own text export (wavelength and value, three decimals): every wavelength equal, every value within half the last printed digit. Two reflectance files from a second depositor (UVProbe 2.71, Chinese-language PC) have no export; their method text names the measuring mode. The corpus test compares every development file that has an export with it (oracle/series_oracle.py --export ... --by-x, corpus/oracle/series/). A file without one is compared only with the same streams read by olefile (BSD-2-Clause) as a black box (--uvprobe-streams), which is not an independent check.

Known gaps

Transmittance, photometric, kinetics and quantitation files have no development file; value codes other than 2 and 4 are returned unnamed. Only the first spectrum of a data set is read. The method pages are kept as text.

Vocabulary (public API of openreadout-spectro, UVProbe)

identifier meaning
UVPROBE_FORMAT_ID, UvProbeReader format id and reader
is_uvprobe, parse, Opened, parsed, values, MAX_DATA_SETS detection, parsing and the parsed file with its y values
HeaderInfo, name, created, organization, software, version, model, series, instrument, serial, header_info, method_page, value_kind, pstr DataSetHeaderInfo and method pages
value_code, data_set, data_set_name, method_pages, header_first_x, header_last_x, header_step, points trace extra and vendor-tree keys
absorbance, reflectance, unsupported_data_set, unknown_value_code, unreadable_header_info, data_set_count, compound_file channel names and check finding codes

How this reader was derived, file by file: provenance log.