ugrep: baixar versions anteriors

What's new?
- update option --encoding=CP437 to retain LF (10) and CR (13) codes in the output #569
- fix a rare edge case for the --encoding option when internally converting a specified code page or a utf-encoding to UTF-8 #568
- update option -z (--decompress) for reasonable xz/lzma decoding limits #566
- update matching mode flags -E, -F, -G, and -P to respect order on the command line to permit overriding (like BSD grep) #562 (see follow-up comments)
- output an error message when regex matching mode flags are used in a config file, instead of silently ignoring them #562 (see follow-up comments)

What's new?
- MinGW portability, tested MinGW-w64 UCRT64 with MSYS2 GCC 16.1; note: make test skips the directory recursive searches due to lack of MinGW symlink handling, see also #553 (apparently, a leading / in a pattern is always expanded to a path when the command line is passed to ugrep's main(), not sure if this can be corrected)
- fail fast when DFA position sets exceed complexity limits #556
Thank you for your contributions ❤️

What's new?
- update regular expression compilation to DFA memory threshold issue #549 (explosive DFA growth is theoretically possible, but should never exceed a worst-case 1GB~2GB memory footprint in ugrep)
- fix man page formatting issue that caused a missing sentence ending #550
- update LZMA SDK to 26.02 #551
Thank you for your suggestions and comments!! ❤️

What's new?
- upgrade to autoconf 2.73 and automake 1.18.1
- update TUI logic to remove terminal() call before print() of selected results, move print() back for more maintainable code (avoid "why is this here, again?")

What's new?
Fixed two issues:
- fix #545 --disable-avx512 tested and working
- fix #544 (same as #537) a bug in the ugrep v7.6 ~ v7.8 (since March 5, 2026) that can't handle very long lines in huge files properly, outputting uninitialized input file buffer contents as if part of the matching line or contents read after the input file buffer as if part of the matching line.
Thank you for your feedback and help with our project ❤️

What's new?
This update includes usability improvements and important updates to protect the user when ug is used to search untrusted file systems:
- improve TUI custom-command option --view to wait for key press after viewing, by specifying + in --view=+COMMAND, such as --view=+file to display file type info in the TUI with CTRL-Y (overrides viewing the file contents) #543
- update TUI to block ANSI OSC escape codes, such as ESC ] P n rr gg bb that may hang the terminal according to Wikipedia
- update the ug+ and ugrep+ scripts to end options in filter commands with --, in case a filename starts with a - (dash)
- improve config parsing to ignore some options that do not make sense to set in a (local) config file
- local config files located in the working directory are only read when owned by the user running the ug or ugrep --config process (already implemented in update 7.7)
Thank you all for your help with the ugrep project ❤️ Stay safe 🙏

What's new?
- support TUI search on slow and incomplete standard input pipes e.g. from tail -f #540
- update options -m (--max-count) and -K (--max-line) to stop reading input upon reaching the stopping criterium, then exit without draining the standard input pipe like GNU grep #538
- improve binary file detection for huge mixed files #537
- add new option ./build.sh --disable-avx512 #534
- local config files located in the working directory are only read when owned by the user running the ugrep process
- fix lzma 7zip api memory leak

What's new?
- new options --max-size and min-size to search files whose physical size is in the specified MIN and/or MAX range
- fix zsh completion syntax error (#528)
- update option --ignore-file to ignore files and directories specified in an .gitignore file as an absolute /glob to ignore those matching the glob under its sub-directories (#526)
- fix emulation of GNU grep option -z (--null-data) to match newlines (zero bytes internally) with pattern \s (space), which requires non-standard regex behavior internally to include matching zero bytes with pattern \s (#525)
- update --ignore-file=FILE to accept a FILE pathname to a non-local gitignore FILE that applies globally to ignore files and directories, similar to --exclude-from=FILE, but with the minor difference that gitignore rules match both files and directories with a single glob (#518 #519 #532)
- fix third-party sourced zopen.c library (BSD open source) one-byte read beyond its allocated struct s_zstate state variable in getcode() (#511)
- fix -m (--max-count) with context options -A or -C sometimes producing garbled after-context output that may cause a crash in the worst case (#510)
- fix reverse sort by date --sort=rchanged and --sort=rcreated not recognized by the TUI at startup (#508)
- update ugrep to search named pipe files specified as arguments on the command line instead of skipping them by default, such as process substitutions; also improve Linux special system files /proc and /sys skipping and/or reading and option -z file read error handling to avoid possible pipe fd leaks when thousands of /proc files are searched that produce (expected) read errors (#503)
- support option --no-empty while using full grep-emulation mode, i.e. when ugrep is renamed to grep
Thank you for your feedback to help improve ugrep ❤️

ugrep: a user-friendly, faster, more capable grep replacement -- search for anything in everything... fast!
What's new?
- smaller and faster executable with extended dual predict-match PM3+PM5 algorithms, see updated benchmarks compiled with Clang for speed and #485
- symlink support for ugrep.exe and ug.exe for Windows #493
- colorization improvement for ugrep.exe and ug.exe for old Windows 7 with ConEmu #491 #495
Thank you for your feedback to help improve ugrep ❤️

What's new?
- fix bug in option -x and regex ^$-anchors used in certain short patterns #489
- mark likely/unlikely branches in hot paths for performance

What's new?
Minor improvements:
- speed up regex pattern search for some more extreme cases #485
- clean up code base to remove shadowed locals
- add safeguards for preventable pattern undermatch; practically impossible to happen with accurate match prediction, but should be guaranteed

What's new?
- fix #486 v7.4 regression issue in the AVX2 SIMD acceleration logic for certain short regex patterns

What's new?
Very pleased to release v7.4 with the following usability enhancements suggested by users:
- more responsive output when searching slow standard input pipes, such as tail -f log | ug pattern #478; this update required several changes to the search engine's acceleration code to support non-blocking standard input, to avoid being too "greedy" for input
- new option argument -t seed7 (or --file-type=seed7) searches seed7 files #480
- faster -g"path/file" (or --glob="path/file". --iglob="path/file") search of the specified directory path glob to search file glob by only recursing into the specified path names/globs; likewise --include-dir="path" is optimized for path globs; these options with paths (globs containing a /) also enable option -r when -d, -r or -R are not specified #481
- clarify the usefulness of the default to not empty-match empty-matching patterns versus GNU grep that does (as with ugrep option -Y) #482
- option -z (or --decompress) now also auto-detects bzip2, XZ, and Zstandard compressed files (i.e. no specific filename extension is required, as is the case already for libz (gzip), compress (Z), zip, and tar/pax/cpio formats) #483
- new option --from=FILE: read additional pathnames of files to search from FILE; this option is useful for example with find ... -print | ugrep --from=- ...' to search specific files found with the findcriteria without usingxargs` #484
- drain the input pipe until EOF when reading non-seekable standard input such as from a pipe, for compatibility with GNU grep, this change re-enables WITH_STDIN_DRAIN in the source code
Updated ugrep v7.4 benchmarks look as good as before
Thank you all for your suggestions and comments!

What's new?
- make options -i (--ignore-case) and -j (--smart-case) respect Unicode letter case conventions #467
- usability improvements

What's new?
- as per request, a minor update to improve search speed when the specified regex pattern is not found in the input when the specified regex starts with a wildcard such as .*, which previously resulted in unnecessary regex match processing overhead

What's new?
- includes fix #463 for v7.2 regression bug in non-SIMD ugrep builds
Notes on unit and regression testing methodology and benchmarking:
- to verify non-SIMD builds I ran 60,000 randomized test patterns each against 100MB text to search and compare to expected matches, this test fully covers the code of the scalar-optimized (non-SIMD) search algorithms (this test was not done for v7.2, which is my fault for thinking the latest change would not impact non-SIMD results, but it did)
- to verify SIMD-optimized builds (SSE2/AVX2/AVX512BW and Neon/AArch64) I ran again 40,000 randomized test patterns each against 100MB text to search and compare to expected matches, this test fully covers the code of the SIMD-optimized search algorithms
- in addition, the RE/flex high-performance regex project also includes many regex pattern test cases
- this release does not affect the search speeds of ugrep v7 i.e. the posted benchmarks for v7 are unchanged

What's new?
- fix a problem with option -i (--ignore-case) that in rare and specific circumstances may not match all of the specified alternate sub-patterns #451.

What's new?
- support the latest bzip3 library upgrades #446 *)
- permit su search of "unreadable" files #447
- do not exit with code 2 when some files can't be decompressed #448
*) please note that./build.sh --with-bzip3 or ./configure --with-bzip3 && make enables bzip3 in ugrep, when libbzip3 is present and found, otherwise ugrep and the indexer skip bz3 files; for help with configuration, run ./build.sh --helpThank you for your feedback to help improve ugrep for everyone!

What's new?
- fix negative character classes when option -i or --ignore-case is used and make ugrep closely emulate GNU grep's character class behavior when options -i and -P are used #445
- appease -Wshadow warnings when building the RE/flex library, the TUI, and indexer (some warnings remain in ugrep.cpp)

What's new?
- add syntax highlighting to the TUI glob editor #443 and other TUI improvements
- add support for option --filter to Windows ug.exe and ugrep.exe #442
- make the ug+ and ugrep+ scripts work if no helper is available #440

What's new?
- fix Windows ugrep.exe TUI issue #438 after TUI usability improvements #435, #436 and #437, which caused some users to struggle with TUI usability issues due to Windows API peculiarities with different Windows OS and different terminal apps

What's new?
- make Windows ugrep.exe TUI cursor keys behave with ConEmu #435
- fix Windows ugrep.exe TUI CTRL-Y file viewing for non-ASCII filenames #436
- minor usability improvements suggested by ugrep users. Thank you for your feedback!

Fix #434 ugrep v7 regression bug: an issue with the regex ^ anchor due to a misaligned begin-of-line position in the buffer after the buffer shifts on large input files. As a result, the -v option did not always work properly on large input files when searching with regex ^ anchors.

What's new?
- faster binary file checking with SIMD (SSE2/AVX2/NEON) UTF-8 validation #421 and benchmarks
- new option --context-separator=SEP
- options -W and -X now also apply to --format fields %o and %O to output hex
- ugrep-indexer option -z (indexes compressed files and zip/7z/tar/cpio/pax archives) no longer indexes hidden directories and files (i.e. "dotfiles") stored in archives, to save space
- fix ugrep-indexer option -z, which was broken due to a code refactoring mistake in a single line of code

What's new?
- new custom output format fields for options --format and --replace #414:
- numeric fields such as %n are padded with spaces when %{width}n is specified with a numeric width > 0
- matching line fields such as %O are cut to width when %{width}O is specified or when %{-width}O is specified to cut from the end of the line
- character context on a matching line before or after a match is output when %{-width}o or %{+width}o is specified for match fields such as %o, where %{width}o without a +/- sign cuts the match to the specified width
- fix bash completions for key bindings #415

What's new?
- option --index always enables --recursive #410
- new bash, fish and zsh ugrep-indexer completions #411
- option --hyperlink outputs absolute search paths #412
- update regex DFA cut optimization that was too agressive for some patterns
- option -o (--only-matching) with -ABC context now also supports multi-line pattern matching with \n (newline) in regex patterns

What's new?
- improved support for legacy grep color environment variable single ANSI color codes format #407, for example: env GREP_COLOR='7;33' ug PATTERN FILE highlights matches in inverted yellow; also ug --colors='7;33' PATTERN FILE does the same and so does ug --colors=iy PATTERN FILE using ugrep's color letter selections
- improved Docker files #408
- improved zsh -ABC options completion #409
- improved option -% (--bool) when multiple -e PATTERN are specified; now aligns with the documentation of option -e to match any of the specified PATTERN (note that -% can be combined with options --and, --andnot, --not. -e amd -N)
- improved option -Q (--query) when combined with options --and, --andnot, --not by activatating option -% (--bool) so that the combined search patterns can be edited as one Boolean query pattern in the TUI
- minor speed improvement when searching in large files with option -f with a file with hundreds, thousands or more words/strings to find (-fwords/4.txt in benchmarks)
- again extensively tested with randomized patterns

What's new?
- improves option -w to support Unicode as planned #234
- makes [[:alpha:]] and other [::] character classes also match Unicode by default, like \p{Class} already does, unless option -U (--ascii or --binary) is used #234
- Unicode word boundaries \<, \>, \b, \B unless option -U (--ascii or --binary) is used #234
- adds static Linux builds for x64 and AArch64 #373 #398 but still needs a script via wget for easy non-admin install on a system (work in progress)
- improved bash and zsh auto-completion usability #395 #394 #392

What's new?
- _quicker!_ 🚀 much faster search speeds on ARM CPUs and also faster overall on other CPUs using essentially the same SIMD algorithms but better optimized #385 see latest ugrep benchmarks for results
- _more fun!_ 🎉 includes ugrep-indexer, which is installed together with ugrep, see ugrep-indexer release notes v1.0.0
Please note
- Distribution maintainers please check if ugrep-indexer and its manual page is indeed also included in your ugrep distribution package and will be installed on user systems. _Thank you!_ 🙏❤️

What's new?
- fix glitch #383 option -A may omit an after context line for regex patterns that start with a wildcard "zero or more" repetition such as .*

What's new?
- revert some of the third-party suggested configure scrip changes to make Homebrew ugrep include compression libraries again #380

What's new?
- updated configure and makefiles to improve SIMD (SSE2/AVX2/AVX512BW/NEON/AArch64) detection and cross-compilation
- removed unused m4/ax_check scripts, adding corresponding PKG_CHECK_MODULES to configure.ac instead #339
- update some of the ctype function calls to avoid potential signedness pitfalls
- AIX build #374
- use statfs() with options --include-fs and --exclude-fs when the preferred statvfs() is not available
- suppress --filter utility stderr messages when ugrep options -s (--no-messages) and -q (--quiet) are used #377
- fix Cygwin --filter utility hanging after fork when worker threads are active, by forcing single-threaded search (a bit crude, but Cygwin fork isn't identical to Unix fork)
- support the NO_COLOR environment variable to suppress color output
- fix TUI scroll glitch (cursor down to scroll screen content up) in non-split screen (glitch appeared when adding the split screen feature)
- TUI line selection mode top row number indicator added, for consistency
- TUI quit & output after line selection mode now behaves the same as non-TUI search output, for consistency
- faster -q (--quiet) search

What's new?
- a minor improvement of the regex syntax to allow escaped spaces #360
- updated POSIX regex lazy quantifier matching in linear time using an advanced DFA transformation algorithm #174

What's new?
- new and updated regex engine to speed up searching patterns described in #288 using a new heuristic DFA cut algorithm
- new option --exclude-fs default to exclude file systems that aren't specified as search targets #349
- new option --all (-@) to #342 to search all files except hidden: cancel previous file and directory search restrictions
- TUI regex syntax highlighting improved for --fixed-strings (-F) when used with option --bool (-% and -%%)
- fix #350 Linux procfs handling not as expected
- fix #355 "context before lines" sometimes incorrectly displayed (due to a misfiring optimization in a recent release)
Notes
- updated benchmarks with four additional cases covered by #288. Many more regex patterns were tested and evaluated on our end, some of which are far more complex than the four cases added to the benchmarks
- further tweaking and tuning of the regex engine to increase performance is possible, but we did not want to hold up releasing 5.0 much longer. Future updates to the regex engine to speed up ugrep further will be considered
- the regex engine updates are machine independent; the machine-dependent parts that use AVX2/AArch64 are located at a lower level in the engine and benefit from the new DFA cut algorithm
- the new DFA cut algorithm does not apply to the fuzzy search regex engine (option -Z) or the PCRE2 library (option -P), which are unchanged
- the new DFA cut algorithm does not apply to index-based search regex patterns with option --index, because it may increase the false positive rate for "normal" patterns. "Leading wildcard patterns" should still be avoided with option --index, such as \w+lookingfor and instead use lookingfor to avoid index-based search false positives (lower performance)

- include PR #343 and PR #344 to update ./build.sh --disable-7zip configure, compilation and linkage
- fix 7zip search failing on 32 bit systems when option -M is used (or -t when setting -M)

What's new?
- 7zip archive search with option -z #185
- apply Unicode normalization to canonicalize literal combining characters in regex patterns #298
- improved TUI TAB directory navigation when searching from the FS root
- updated ugrep.exe option -P to use PCRE2 latest stable version 10.42
Future plans (work in progress)
- implement a new DFA-based regex search optimization for patterns described in #288
- speed improvements for Linux with io_uring (could be 2x faster)
- update ugrep benchmarks when some or all speed improvements are included (no update yet, because v4.5.0 is as fast as the current benchmarks show)

What's new?
- fix an issue with option -Zbest fuzzy matching

What's new?
- updated option --index for accelerated index-based search, now also indexes and searches archives and compressed files with option -z
- new ugrep.exe with option --index for accelerated index-based search on Windows
- updated ugrep-indexer project to support archive/compressed file indexing with a new ugrep-indexer.exe version for Windows
- updated MacOS moar-pager #328
- faster search with -ABC context options #329
Future plans
- a major 4.4 or 5.0 release with performance enhancement #288 by updating the regex engine to advance fast to a partial match and then "look backward" for a full match
- make the ugrep-indexer command part of the ugrep tools when the ugrep-indexer project graduates to stable (ugrep-indexer is currently 0.9.2 beta)
- Unicode searching with Unicode pattern normalizing #298
- other additions and improvements

What's new?
- the ug command no longer quits with an error message when no default .ugrep config file was found
- bzip3 is supported but not required to build ugrep, see also #316; to use bzip3 with ugrep, execute ./build.sh --with-bzip3.
- fix the output of + separators by no longer using them #317 for GNU grep compatibility
- fix option -v with -ABC context #319
- fix configuration file option arguments that may got lost and causes option argument errors in some cases after parsing a config file, such as colors=
- allow config file importing in config files using config=FILE (but does not permit recursive imports) see also #320
- add Dockerfile-minimized #322
- minor usability improvements

What's new?
- added brotli compressed files search support #312, see notes below
- added bzip3 compressed files search support #311, see notes below
- updated option --pretty to accept an optional WHEN argument never, always, auto similar to the --color option
- improved TUI screen blanking response when a new search starts, while also avoiding screen flicker
- fix #313 option --zmax>1 when a decompression library produces a decompression error when decompressing a nested compressed file
- PCRE2 to enable option -P
- zlib to enable option -z
- bzip2 (also sometimes used by the zip format)
- lzma/xz (also sometimes used by the zip format)
- lz4
- zstd
- brotli (new optional dependency, not included in ugrep.exe)
- bzip3

- new official ugrep.com web site with the ugrep user guide and installation instructions, this site will always be safe and guaranteed to be ad free and cookie free; ugrep.org forwards to ugrep.com
- double short option -%% enables both --files --bool (a single -% still enables only --bool) for convenience
- updated thread pool scheduling and execution with thread affinity and priority settings, implements #305
- improvements and fixes for minor (mostly cosmetic) issues
- updated ugrep v4.3.2 benchmarks.

- TUI Tab directory navigation now also works when file and directory arguments are present on the command line
- TUI ctrl-Y now also supports viewing standard input when piped to ugrep
- TUI ctrl-Y now also supports viewing files stored in zip/tar/pax/cpio archives, including nested compressed files and archives
- TUI regex syntax highlighting with user-definable colors; the highlighting is sensitive to the current regex syntax, i.e. ERE (default), BRE (-G), --bool, and -F (--fixed-strings).

- new TUI split screen, activated with CTRL-T, and new option --split to start the TUI with a split screen (add split to your .ugrep configuration file to enable) #254
- improved TUI screen updating
- refactored the search engine internals to keep the search engine alive until ugrep terminates to optimize the TUI matching file preview pane display speed
- improved option -o (--only-matching) layout when context options -ABC are used
- fixed --zmax=n for n>1 that had a low chance to hang as discovered recently (about once in a few thousand runs) when searching (deeply) nested zip or tar archives
- fixed compressed file search of files that are exactly one byte long, instead of rejecting those files as being empty
- updated performance benchmark results: ugrep is faster than other grep tools on x64 and arm64 machines in nearly every test of 564 benchmark test cases

- improved TUI to display search progress and final stats in a status line at the bottom
- improved TUI ctrl-Y to jump to the corresponding line number in the pager #296
- improved recursive search performance #297
- improved compatibility with GNU grep #273
- improved detection of special devices to avoid searching them