Releases: ilri/csv-metadata-quality
Releases · ilri/csv-metadata-quality
Version 0.6.1
Fixed
- Missing region check should ignore subregion field, if it exists
Changed
- Use SPDX license data from SPDX themselves instead of spdx-license-list because it is deprecated and outdated
- Require Python 3.9+
- Don't run
fix.separators()
on title or abstract fields - Don't run whitespace or newline fixes on abstract fields
- Ignore some common non-SPDX licenses
- Ignore
__description
suffix in filenames meant for SAFBuilder when checking for uncommon file extensions
Updated
- Python dependencies
Version 0.6.0
Changed
- Perform fix for "unnecessary" Unicode characters after we try to fix encoding issues with ftfy
- ftfy heuristics to use
is_bad()
instead ofsequence_weirdness()
- ftfy
fix_text()
to not change “smart quotes” to "ASCII quotes"
Updated
- Python dependencies
- Metadata field exclude logic
Added
- Ability to drop invalid AGROVOC values with
-d
when checking AGROVOC values with-a <field.name>
- Ability to add missing UN M.49 regions when both country and region columns are present. Enable with
-u
(unsafe fixes) for now.
Removed
- Support for reading Excel files (both
.xls
and.xlsx
) as it was completely untested
Version 0.5.0
Added
- Ability to check for, and fix, "mojibake" characters using ftfy
- Ability to check if the item's title exists in the citation
- Ability to check if an item has countries, but no matching regions (only suggests missing regions if there is a region field in the CSV)
Updated
- Python dependencies
Fixed
- Regular expression to match all citation fields (dc.identifier.citation as well as dcterms.bibliographicCitation) in
experimental.correct_language()
- Regular expression to match dc.title and dcterms.title, but ignore dc.title.alternative
check.duplicate_items()
- Missing field name in
fix.newlines()
output
Version 0.4.7
Changed
- Fixing invalid multi-value separators like
|
and|||
is no longer classified as "unsafe" as I have yet to see a case where this was intentional - Not user visible, but now checks only print a warning to the screen instead of returning a value and re-writing the DataFrame, which should be faster and use less memory
Added
- Configurable directory for AGROVOC requests cache (to allow running the web version from Google App Engine where we can only write to /tmp)
- Ability to check for duplicate items in the data set (uses a combination of the title, type, and date issued to determine uniqueness)
Removed
- Checks for invalid and unnecessary multi-value separators because now I fix them whenever I see them, so there is no need to have checks for them
Updated
- Run
poetry update
to update project dependencies
Version 0.4.6
Added
- Validation of
dcterms.license
field against SPDX license identifiers
Changed
- Use DCTERMS fields where possible in
data/test.csv
Updated
- Run
poetry update
to update project dependencies
Fixed
- Output for all fixes should be green, because it is good
Version 0.4.5
Added
- Check dates in dcterms.issued field as well, not just fields that have the word "date" in them
Updated
- Run
poetry update
to update project dependencies
Version 0.4.4
Added
- Accept dates formatted in ISO 8601 extended with combined date and time, for example: 2020-08-31T11:04:56Z
- Colorized output: red for errors, yellow for warnings and information, green for changes
Updated
- Run
poetry update
to update project dependencies
Version 0.4.3
Changed
- Reformat with black
- Now requires Python 3.7+ for pandas 1.2.0
Updated
- Run
poetry update
to update dependencies - Expand check/fix for multi-value separators to include metadata with invalid separators at the end, for example
Kenya||Tanzania||
Version 0.4.2
Changed
- Add field name to the output for more fixes and checks to help identify where the error is
- Minor optimizations to AGROVOC subject lookup
- Use Poetry instead of Pipenv
Updated
- Update python dependencies to latest versions
Version 0.4.1
Changed
- Reduce minimum Python version to 3.6 by working around the
is_normalized()
that only works in Python >= 3.8