- Some bug fixes for
UnicodeSet
.
- Upgraded dependencies to
@eslint-community/regexpp
v4.8.0 andrefa
v0.12.0. is{Potentially,}{Empty,ZeroLength}
,isEmptyBackreference
,getLengthRange
, andisLengthRangeMinZero
now require a flags argument.- Changed the semantics of
matchesAllCharacters
.
- Added support for the
v
flag.toUnicodeSet
is the newtoCharSet
.hasStrings
will return whether a character element contains strings.
getConsumedChars
will return all characters consumed by an element.Chars.maxChar(flags)
will return the maximum character (either0xFFFF
or0x10FFFF
) depending on the given flags.
- Upgraded dependencies to
@eslint-community/regexpp
v4.5.0 andrefa
v0.11.0.
getClosestAncestor
now supports any number of arguments.
- Fixed
getLongestPrefix
not caching correctly.
getLengthRange
will now throw aRangeError
instead of returningundefined
for empty arrays.
- Added
isLengthRangeMinZero
function that is equivalent togetLengthRange(e).min === 0
but implemented more efficiently. - Added
onlyInside
option forgetLongest
function.
canReorder
can now prove more alternatives to be safe to reorder.canReorder
will now create a cache if none is provided.
- Dropped support for NodeJS 10. The library will likely still work on this NodeJS version, but it won't be tested and NodeJS-10-specific bugs will be rejected.
- Added
containsCapturingGroup
. - Added
getLongestPrefix
. - Added
canReorder
.
- The
FirstLookChar
,FirstFullyConsumedChar
, andFirstPartiallyConsumedChar
interfaces are now immutable.
- Added transparent caching for all functions taking flags.
- Added
FirstConsumedChars
andFirstLookChars
namespaces. They contain methods for working withFirstLookChar
s andFirstConsumedChar
s. - Added
FollowOperations.continueOutside
to improve analysis inside lookaround assertions. - Added
FollowEndReason
type. followPaths
now supports alternatives as the start element.- The
getFirst{Consumed,}Char*
functions now supports alternatives as the after-this element.
- The
getFirst{Consumed,}Char*
functions can now look outside of lookarounds to return better results. - Improved how the
exact
property ofFirstConsumedChar
s is determined.getFirstConsumedChar
will now return better results. - Improved documentation.
- Lots of internal refactoring and improvements.
- Fixed that
getFirstConsumedChar
sometimes returned partially consumed chars with trivially rejecting looks instead of a fully consumed char.
- Fixed
getFirst{Consumed,}Char*
functions taking exponential time for some word boundary assertions.
- Updated refa and regexpp.
- All
getFirst{Consumed,}Char*
functions can now handle word boundary assertions.
- Added support for
d
flag.
- Private types aren't exported anymore. (They were never supposed to be exported in the first place.)
- Updated refa to v0.8.
isEmptyBackreference
: Fixed stack overflow for circular nested backreferences.
is{Empty,Strict}Backreference
: More efficient implementation.hasSome{Ancestor,Descendant}
: Node can now be given to the functions instead of condition functions.getClosestAncestor
: The return type is now stricter and exported asClosestAncestor<A, B>
.
is{Empty,Strict}Backreference
: These two functions will now properly account for capturing groups inside negated lookarounds.
- Package meta information.
Initial release.