ugrep v3.12.3

Speed improvements to further speed up common search patterns. Improved man page. The ug and ug+ commands now enable --pretty by default (that is, if a .ugrep configuration file does not override it with no-pretty). Fixed a potential pattern match issue in the new v3.12.2 optimizations. Working on additional SIMD optimizations with SSE/AVX and neon/AArch64 to make ugrep even faster. Correctness is most important, so optimizations are incrementally released and extensively tested.

ugrep v3.12.2

Speed improvements to further speed up common search patterns. Improved man page. The ug and ug+ commands now enable --pretty by default (that is, if a .ugrep configuration file does not override it with no-pretty).

ugrep v3.12.1

New ug+ and ugrep+ commands to search pdfs, documents and image metadata when the corresponding tools are installed, includes a runtime check for tool availability; improved --filter option; speed improvements.

Ugrep runs faster than grep and ripgrep on Apple Silicon M1 and x64 machines for most common searches, such as searching for a few words to tens and hundreds of words in large files. It depends on the available hardware and the complexity of the search patterns.

Experiments on a MacBook PRO M1. The average elapsed time of 100 test runs of ugrep, ripgrep and GNU grep with increasing number of randomly picked words from the enwik8 100MB benchmark file:

Results are for ugrep -on -f words enwik8 to search with line numbers and option -o to reduce output overhead. The performance of ugrep is better than ripgrep and better than GNU grep for smaller sets of strings to search. However, ugrep and ripgrep are more or less comparable when searching more strings and when searching more complex regex patterns. It depends.

ugrep v3.12.0

New ug+ and ugrep+ commands to search pdfs, documents and image metadata when the corresponding tools are installed; improved --filter option; speed improvements.

Ugrep runs faster than grep and ripgrep on Apple Silicon M1 and x64 machines for most common searches, such as searching for a few words to tens and hundreds of words in large files. It depends on the available hardware and the complexity of the search patterns.

ugrep v3.11.2

This update includes usability improvements. We improved option --hyperlink to optionally take parameters PREFIX and +. The PREFIX parameter replaces file:// in the link with a custom PREFIX://. The + included the line number in the link, and when ugrep option -k is used, also includes the column number in the link.

ugrep v3.11.1

Smarter interactive navigation in the TUI when option --tree is used with option -l or -c, enabling the navigation keys CTRL-S (jump forward), CTRL-W jump backward), TAB (chdir into directory and save query state), SHIFT-TAB (chdir .. and restore query state), CTRL-Y (view file).

ugrep v3.11.0

Updated --format output when used with option -o by also enabling option -u to show all matches automatically; improved --json, --xml, --csv output when used with option -o; updated option -o output with headings (when applicable) for every match like GNU grep; updated options --format and --replace field %m to output the number of matches sequentially; new --format and --replace field %M outputs number of matching lines.

ugrep v3.10.1

Fixed a performance issue with specific regex patterns when used with case-insensitive pattern matching.

ugrep v3.10.0

New option --tree to output directory trees of files for options -l (--files-with-matches), -L (--files-without-match) and -c (--count). Updated option --pretty to enable --tree when output is sent to a terminal. Disable with --no-tree to output a traditional grep flat list.

For example, when pretty is specified in a .ugrep file, ug -l (--files-with-matches) outputs directories in a tree-like format similar to the unix tree utility:

ugrep v3.9.7

A minor update to handle gnutar archive extension field formats to search huge files (>4GB) stored in (gnu) tar archives.