Skip to content

Latest commit

 

History

History
238 lines (170 loc) · 10.9 KB

NEWS.md

File metadata and controls

238 lines (170 loc) · 10.9 KB

opusreader2 0.6.3 (2024-03-12)

Bug fixes.

  • fix sample and time metadata parsing for read_opus(dsn, data_only = TRUE). Previously, extraction of the timestamp failed and data extracted errored in that case, because there was the required "history" and "sample" blocks weren't extracted temporarily before, as required. Now, read_opus(dsn, data_only = TRUE) successfully extracts an extra element basic_metadata, as it does for data_only = FALSE (the default). This extra information does unlikely to break existing pipeline code that extracts spectra with {opusreader2}, because it is a nested list element. This patch release also resolves a warning when parsing time information, that was due to an extra tab ("\t") that was present in the history text for specific files. Thanks @mtalluto for the fix. Added extra tests to check for errors and warnings in the example files for both data_only = FALSE and data_only = TRUE). Thanks to @dylanbeaudette and @esteveze for reporting the failing extraction of metadata. Issue report: #104. PR fixed: #105.

opusreader2 0.6.2.9000 (2023-12-27)

  • Select 5 OPUS binary files from different instrument types for tests and vignette ((#103)
  • Update first part of vignette for CRAN (#103).

opusreader2 0.6.2 (2023-12-23)

Bug fixes

  • Hotfix for commit c8ff2cd, which accidentally caused a regression, leading to issues #101 and #102. It was unnoticed but could have been diagnosed with the {testthat} tests in place. There was also an update of the {tic} template, which did not invoke tests yet in the continous integration (CI) run (passed because of unconfigured tests). When restricting relevant export, roxygen2 @export tags were removed and @internal added for calc_parameter_chunk_size(), which made those functions unavailable even internally ("Error in UseMethod("calc_parameter_chunk_size") : no applicable method for 'calc_parameter_chunk_size' applied to an object of class "parameter")"

Documentation

  • read_opus(): in return element ab, state Log10 explicitly for calculating apparent absorbance (#94; @zecoljls).
  • Only export functions relevant to users:
    • core: read_opus(), read_opus_single()
    • S3 methods for calc_parameter_chunk_size()
    • helpers: opus_file()

opusreader2 0.6.1 (2023-11-12)

OPUS data support

  • Support quality test report (#81). This block can be found in ./inst/extdata/new_data/issue81_A1.1.0. read_opus() returns this block as "quality_test_report" in the list output.

opusreader2 0.6.0 (2023-11-12)

  • Add first unit tests using the {testhat} framework.
  • Allow non-parsable blocks. Add new default so that all blocks that not yet mapped are showing up as warnings instead of an error. These blocks will be named as "unknown" elements in the output of the read_opus() list.

OPUS data support

  • Internal refactoring (see below) fixes two reading issues:
    • ./inst/extdata/new_data/issue94_RT_01_1_23-02-21_13-23-54.0: from Bruker 2023 Alpha II mid-IR spectrometer. Due to internal refactoring of header parsing (see below) (#94)
    • ./inst/extdata/new_data/issue82_Opus_test: from Bruker MPA FT-IR spectrometer. Parse block "b0-c0-t144-a1", text type 144 with special offset in parse_chunk.parameter(). For now classify this block as block type "report_unknown" (waiting finalize naming until confirmed with screenshots from the Bruker OPUS sofware). Also fix time_saved by not relying on language settings (#82)

Internal refactoring

  • Simplify header parsing in parse_header().
  • Work with raw vectors instead of connection objects to read binary data. Parse raw vectors directly for functions in read_bin_types() and use subsetting to slice raw vectors in base::readBin() calls instead instead of seek(), which was used previously to reposition cursors in raw connections.
  • get_meta_timestamp(): omit language dependent logic using "time saved" regular expressions for matching time saved from history block. The first time of sorted POSIXct candidates will be returned as time saved.

opusreader2 0.5.0.9000 (2023-06-05)

  • implement a basic_metadata list element for "opusreader2" class containing key metadata (#85)

opusreader2 0.5.0 (2023-06-03)

  • Name first level of list (class "list_opusreader2") with base file name of given data source name (DSN) (#83)

  • Fix "list_opusreader2" indenting when reading files in parallel (#80)

  • Add support for progress bars in read_opus() (#75)

  • Introduce type-stable classes for read_opus() and read_opus_single() output (#72):

    • classes "list_opusreader2" and "opusreader2"

opusreader2 0.4.1 (2023-03-19)

  • patch when read_opus(..., parallel = TRUE): unlist resulting list one level (chunk level); #80.

opusreader2 0.4.0 (2023-03-14)

  • Feature progress bar for read_opus() when reading multiple files in parallel #75.

opusreader2 0.3.0 (2023-02-16)

  • The exported functions are now (#74):
    • read_opus(): Read one or more OPUS files from data source name (dsn)
    • read_opus_single(): Read a single OPUS file
    • calc_parameter_chunk_size(): Calculate the parameter chunk size in bytes

opusreader2 0.2.0 (2023-02-15)

  • Introduce new S3 classes for the main functions exported (#72):
    • read_opus(): S3 class c("list_opusreader2", "list")
    • read_opus_single(): S3 class c("opusreader2", "list")

opusreader2 0.1.0 (2023-02-08)

Refactoring

  • Internal refactoring (R/create_dataset.R). Implement a new key-value mapping logic for assigning the integer coded header information. The new order in the (composite) key strings follows the sequence of block, channel, text and additional type information. The better line-by-line layout of composite keys and mapped information types simplifies the detection of new kind of spectral data and parameters that are encoded in header entries (#60).

  • Introduce consistent and proactive error reporting when a composite key in are not yet mapped because they are not yet known (R/create_dataset.R). This error message includes a recipe how to report new OPUS files with yet unsupported block types (i.e. new instrument features) for {opusreader2}. Together with the composite key generated from the respective the header entry, a step-by-step reporting as GitHub issue is proposed. (#60)

opusreader2 0.0.0.9002 (2022-12-23)

Documentation

  • Update return value of parsed OPUS spectral blocks in parse_opus()

opusreader2 0.0.0.9001 (2022-12-18)

Start versioning with {fledge}.

spectral-cockpit.com proudly introduces {opusreader2} to read binary files from FT-IR devices from Bruker Optics GmbH & Co in R. It is a powerhouse that fuels speedy extract-transform-load (ETL) data pipelines in spectroscopy applications. You can continue using state-of-the-art commercial devices for what they are good at: measurements. Meanwhile, you can rely on open source technology and trans-disciplinary knowledge to design data processes, and make best use of the spectroscopic source of information.

{opusreader2} parses and decodes the at first glance puzzling file header first. The implementation then uses this mapped information as a recipe to read particular data types from different blocks. Specific byte chunks to be interpreted are defined by position (offset), read length, bytes per element, and type (e.g., string, float). With this, all the data can be read and parsed. We mitigate lock-in at file level. Hence we foster reproducible and trustworthy processes in spectral workflows. Nowadays, the new business logic is being more and more transparent in code, methods used and services offered. Tightly link and make input data, metadata and outcomes available for economical scaling-up of diagnostics.

  • Extract, transform and load data directly from OPUS binary files

Providing the data and metadata from measurements connects downstream tasks in order to make IR spectroscopy a ready-made, automatec for diagnostics and monitoring (platform):

  • Quality control of measurements; monitoring workflow and metadata
  • Continuous spectroscopic diagnostics (data processing, model development, inspection, adaption, prediction, and validation). Use MLOps principles.

With our package you can directly read and parse from binary files without compromising a single bit of precious information saved in these filled OPUS binary files.

read_opus() is the main function exposed that reads and parses OPUS binary files from various data sources names (dsn). Currently, we support the following dsn types:

  • files(s): character vector with one path to OPUS file or multiple paths to individual OPUS files
  • folder: character of length 1 with path to folder with OPUS files to be read recursively. Only reads OPUS files with .<integer> extension (Usually starting from .0 for unique sample names per measurement.

File names of OPUS files can possibly include plate positions that are postfixed to the sample names. This is an option in OPUSLab. Kindly note that the associated metadata (sample name/ID) and plate position are also stored internally so that file name changes after measurement could be tracked.

read_opus offers four arguments:

  • dsn: data source name
  • data_only: switch to extract only spectral data blocks without additional information like measurement parameters or environmental conditions.
  • parallel: not enabled by default. Speed up reads of 1000s of files by chunking list of files across parallel workers. Cross-platform via unified {future} framework in R.
  • progress_bar: optionally show interactive progress bar for single-threaded or asynchronous reads.

The interface is minimal and the job of the generic reader function is well defined by design. This is to make maintenance easy and to avoid breaking changes in future releases of the package. We importantly avoid feature overload like this. We plan to release specific helper and wrapper functions that can come in handy for tailored uses and diagnostic environments. They may also extract or post-process spectroscopic data and metadata pipelines. Check out more soon in future releases.