• csvtk v0.39.0
  • csvtk:
    • improve performance.
    • prevent overwriting input and auxiliary files.
    • report file I/O and compression errors, even with -I/--ignore-illegal-row.
    • fix column range selection.
    • prevent integer overflow when expanding column ranges.
    • avoid repeating duplicate-named columns during exclusions and fuzzy selection.
    • accept literal column names containing regular-expression characters.
    • make --immediate-output a global flag and flush both CSV and file output buffers.
    • preserve explicit stdin input when using -X/--infile-list.
    • infer column counts from retained records when skipping empty or illegal rows, and skip empty rows with differing widths under -E.
  • csvtk add-header:
    • reject mismatched column counts in subsequent input files.
  • csvtk cat:
    • preserve final lines without a newline and fix progress reporting and parameter validation.
  • csvtk comb:
    • fix reading of multiline fields, long records, comments, and many input files.
    • reduce memory usage when generating combinations.
  • csvtk comma:
    • preserve numeric precision and scientific notation.
  • csvtk concat:
    • report ambiguous column names.
  • csvtk corr:
    • reject multiple input files instead of silently ignoring extra files.
  • csvtk csv2json:
    • fix JSON escaping, numeric formatting, and trailing commas with -K/--skip-key.
    • require exactly one resolved key column.
  • csvtk csv2md:
    • escape literal pipes and render cell line breaks without breaking the table.
  • csvtk csv2rst:
    • preserve multiline cells and align borders with Unicode padding.
  • csvtk csv2xlsx:
    • fix data loss in wide tables.
    • generate distinct, valid sheet names for input files with conflicting names.
  • csvtk cut:
    • fix a panic when combining -H, -m, and excluded columns.
  • csvtk filter:
    • fix --any when selected fields contain non-numeric values.
  • csvtk filter2/mutate2/mutate3:
    • fix numeric precision and string handling, and allow mixing $x and ${x} references.
  • csvtk fix:
    • honor --na when padding rows with -n/--buf-rows.
  • csvtk fix-quotes:
    • preserve valid escaped quotes and support regular-expression characters as delimiters.
  • csvtk fix-quotes/space2tab:
    • reject invalid and overflowing buffer sizes before opening output files.
  • csvtk fold:
    • honor --ignore-case, retaining the first spelling of each key.
  • csvtk head/ncol/headers/space2tab/dim:
    • fix failures when processing many input files (with --cols for dim).
  • csvtk inter:
    • fix intersections with empty inputs.
  • csvtk join:
    • fix joins with header-only inputs.
    • fix misplaced keys and missing values in outer joins with reordered key fields.
    • retain generated column names after an empty intermediate join and reject repeated key columns.
  • csvtk long2matrix/matrix2long:
    • validate value bounds and treat NaN as non-numeric when filtering.
  • csvtk mutate:
    • fix column insertion with --remove and validate capturing groups.
  • csvtk plot:
    • reject invalid field selections, non-finite dimensions, and invalid axis limits.
    • accept uppercase output format names.
  • csvtk plot bar:
    • fix grouped bar alignment and reject duplicate categories within a group.
    • honor X-axis limits and report unsupported logarithmic scales.
  • csvtk plot box/line:
    • pad large constant values without losing precision.
  • csvtk plot box/line/hist:
    • validate logarithmic domains and handle constant positive ranges before opening output files.
    • reject non-finite axis ranges before rendering.
  • csvtk plot hist:
    • fix the MEAN label with --percentiles.
    • support logarithmic counts with empty bins, reject non-finite values, and report empty samples before computing percentiles.
    • reject unrepresentable bin widths and ranges before rendering.
  • csvtk plot line:
    • fix categorical axis ticks for single categories and custom axis limits.
  • csvtk pretty:
    • fix named alignment with row numbers and reject out-of-range alignment fields.
    • validate alignment range bounds before expanding them.
  • csvtk replace:
    • fix group ranges and fuzzy selectors modifying unintended columns.
  • csvtk replace/fold/spread:
    • resolve field exclusions independently for each selector.
  • csvtk replace/rename2:
    • preserve literal dollar signs in key-value replacements.
  • csvtk round:
    • fix rounding of integers, decimals, and scientific notation.
  • csvtk sample:
    • reject NaN sampling proportions.
  • csvtk sep:
    • honor case-insensitive literal separators and keep output names and widths consistent when a named field overrides -H.
  • csvtk sep/mutate/watch:
    • require exactly one selected field, including after range and exclusion expansion.
  • csvtk sort:
    • fix default sorting of single-column input and open ranges selecting only the last column.
    • correctly sort by numeric-looking column names and negative open ranges.
  • csvtk split:
    • reject unsafe output directories before deleting files, even with --force.
    • honor --buf-rows 1 and fix appending output when reading from stdin.
    • limit open files and memory when splitting by chunk count.
  • csvtk split/csv2xlsx/splitxlsx:
    • reject generated output filenames that overwrite input files.
  • csvtk splitxlsx:
    • fix panics on short or blank rows.
    • fix open field ranges and fuzzy field matching.
    • honor -o/--out-file when listing sheets and keep the workbook open until processing finishes.
  • csvtk summary:
    • fix field ranges with the default count operation and honor --rand-seed.
    • resolve data and group selectors independently, including exclusions, and reject overlapping columns.
    • reject overflowing field numbers.
    • avoid overflow when computing medians and quartiles of finite values.
  • csvtk summary/spread/sep/rename/rename2/csv2rst/corr/watch/pretty/fix/del-quotes/split/splitxlsx/space2tab/fix-quotes/transpose:
    • honor -U/--delete-header.
  • csvtk watch:
    • prevent image output from overwriting inputs or tabular output.
    • reject invalid report frequencies and handle inputs without finite numeric values.
    • preserve non-numeric and non-finite records with --pass.
  • csvtk xlsx2csv:
    • honor -o/--out-file when listing sheets.