reflex v4.2.1

What's new?

• minor update to adjust the current input pointer by one when not matching anything

reflex v4.2.0

What's new?

• new pkg-config files reflex.pc and reflexmin.pc and updated configure and makefiles
• minor update of reflex command option -p to permit empty lexer specifications #202 (*)

*) performance profiling statistics are collected for lexer rules execution, i.e. when no rules are specified no meaningful performance statistics are collected.

reflex v4.1.2

What's new?

• updated configure scripts with improved cross compilation and improved SIMD CPU extension detection
• cast negative ctype function arguments (problem detected on NetBSD 10 #198)
• verified on ARM6 (Debian), ARM7 (Debian), AArch64 (MacOS M1, Android termux), x86_64 (MacOS Intel, Cygwin, Debian)

reflex v4.1.1

What's new?

A minor update to correct a DFA construction problem for POSIX regex lazy quantifiers matching too much in some cases

reflex v4.1.0

What's new?

• improved lazy quantifiers for POSIX regex lazy matching in linear time (i.e. no backtracking) using an advanced DFA transformation algorithm first introduced in RE/flex in 2016 and extended in this release
• add option --[no]yypanic for Flex++ compatibility

reflex v4.0.1

What's new?

• improved option %o find
• two new examples, including an AWK-like fast search demo in C++, to demonstrate option %o find
• fix #195

reflex v4.0

What's new?

• faster Matcher::find() with a new DFA cut algorithm to optimize match prediction speed and accuracy, see also ugrep 5.0
• apply Unicode pattern canonicalization with reflex::convert(..., reflex::convert_flag::unicode)

reflex v3.5.1

Minor improvements.

reflex v3.5.0

• updated to Unicode 15.1
• update \X to match Unicode characters
• clarify . (dot) in the documentation when used with %unicode-enabled lexers, which is a catch-all pattern e.g. to report lexical errors

reflex v3.4.1

Word boundary anchors (\<, \>, \b and \B) can now be used anywhere in a pattern with the default RE/flex regex engine. Backtracking on word boundary anchors is performed by the regex engine to match input when anchors are embedded in a regex pattern (not at the ends of a pattern) such as bar.*\bfoo.

However, no backtracking is performed by the fast directly-coded DFA engines generated with --fast, which means that word boundary anchors that require backtracking, such as bar.*\bfoo, may not match with --fast. Please note that tokenizers and scanners typically don't use word boundaries in their patterns, but this new feature was added as part of a general RE/flex regex pattern library update and is also supported by its fuzzy matcher variant.