-chore Update deps from @0323pin and includes fix for clippy lint
- feat Update deps from @0323pin
- Bugfix Fixed output column reordering
- Updated deps and added dependabot
- [Bugfix] Invalid thirdparty file fixed
- [Improvement] Updated deps to get latest gzp and thereby latest flate2, fix thirdparty.yml
- [Improvement] Updated deps to get latest gzp and thereby latest flate2
- Improvement Update from structopt to clap v3
- [Improvement] Updated deps
- bugfix Apply unescaping to headers when selecting via
-F
- Improvement Allow for passing escaped sequences in for the input and output delimiteres. i.e '\t' or '\n' can be passed in directly instead of needing to add $'\n' on the command line.
- Improvement Add a flag
-I/--use-input-delimto reuse the input delimiter if it is a literal as the output delimiter. An error will be raised if-Lis not specified or if-Dis specified.
- Bugfix allow arguments of options to start with a "-".
- Performance improvements. On literal byte separators
hckis now faster than it previously was and is much faster on inputs that only use the first few fields in a line.
- Bugfix for files using fast-path code that have empty first lines
- Added thirdparty license in prep for conda-forge
- Change the number of compression threads used by default
- Update gzp version
- Output BGZF output by default, which can be indexed and queried with tabix
- Read input gzipped files a
MultiGzDecoderwhich is more flexible
- PGO build fix
- Speed up edge case where columns may have already been consumed i.e.
-f1-3,5,2- - Remove pigz binary from compression search
- Add native gz decompression
- Add native gz output compression via
gzp
- Bugfix Better handling of duplicate selected fields, fixed output ordering when duplicate fields were selected. Added clarification to README regarding mixing by-index and by-header field selction / reordering.
- Bugfix, allow headers specified to be excluded to not be found
- PR24 Removed the now defunct profile guided optimization shell scripts and all references to them in favor of the
justfilethat was added inv0.5.0 - Bugfix fixes incorrect handling of header line for non-stdin inputs, fixes incorrect parsing of last header fields (now strips newline before matching), fixes option parsing so that the
-Fand-Eoptions wont' try to consume the positional input arguments. Huge thanks to @learnbyexample for their detailed bug report. - Change: An error will now be raised when a specified header is not found. This differs from the convention used by the selecion-by-index, which tries to match
cut. The reasoning is that it is generally harder to type out each header field and if a header is not found you want to know about it.
- Fix the version in the binary to match the actual version
- Added
--exclude|-eflag to select a set of fields to exclude. These fields may overlap with the-fflag and take precedence over fields selected by-f. - Added
--exclude-header|-Eflag to select a set of headers to exclude. These may mix and match with-e-fandF. The-rflag will cause the headers to be treated as a regex. - As part of the
-eadditional, the default behavior if now headers or fields are specified is to assume-f1-, which allow the user to dohck -e 3,8,290. - pigz is now a supported decompression binary, if it's not present
hckdefaults back togzip. - Decided against adding a greedy heuristic because it actually had worse performance on the most common case of
\s(but better on[[:space:]], which was odd).- The place where this would make sense would be searching a literal space character greedily (like awk), but that kind of goes against the way the delimiters are documented to work
- It may be worth adding that special case at some point?
- Moves CI to using justfile instead of pgo scripts.
- Fixes several issues in benchmarks
choosewas not using fastest path and had the wrong input file- All tools splitting by a space regex were incorrectly parsing the header line in the multichar data, header line is now fixed
- Fixes bug with reordering of only two fields
- Fixed deb CI