1 unstable release
0.0.0 | Jan 30, 2022 |
---|
#69 in #focus
52KB
890 lines
nytri
— Not Your Typical Regex Implementation
nytri
is a low-level Regular Expressions library with a focus on user-control.
What does this mean? In practice, it means that nytri
is designe to provide its users an unusual
level of access to the underlying Regex engine. Things like:
- Stateful matching — i.e. feeding pieces of a match piece-by-piece
- Saving & restoring match state
- ... and many more planned.
This library is under initial development. It should not be used in any serious setting, and does not yet come close to the feature set that should be expected by a quality regex implementation.
Current & Planned features
Before reading this list, note: nytri
will always guarantee linear-time matching. Because of this,
some features are not possible to implement fully (like backreferences).
Current:
- (very) Basic full-text matching (grouping w/ parens, alternatives, literals)
- Stateful matching (via
Matcher::feed
&Matcher::is_match
) - Matcher state save & restore (via
Matcher::get_state
&Matcher::from_state
)
Planned:
- Error display options & configuration
- Syntax:
- All of:
.
,*
,+
, and?
— with the standard meanings. - Character escapes
- (ASCII) Character classes
- Unicode character classes
- All of:
- Matcher configuration — will be clarified in the future
- Extracting capture groups (Named and unnamed)
- Possible: intentionally don't capture certain groups, for efficiency.
- Possessive capture groups
- Lookahead & lookbehind (plus their negative variants)
- Generic atom and text types (e.g. alternatives to
char
and&str
, or matching on&[u8]
implemented as "just another text type") - Limited support for backreferences (there is a useful subset that can be matched in linear-time)
Dependencies
~290–750KB
~18K SLoC