Commit Graph

18 Commits

Author SHA1 Message Date
1db61108e4 Allow pipes that have a missing operand - if an operand is missing, it is replaced with a zeroLengthMatchState(), which always has a zero-length match 2025-01-25 22:36:58 -05:00
08e01a1c81 Loosened restrictions for concatenation - It's okay if one of the
elements is missing
2025-01-25 13:09:47 -05:00
93903fc557 Allowed creation of empty non-capturing groups 2025-01-25 13:04:36 -05:00
4966a222f9 Added detection of empty parentheses, as zero-length matches 2025-01-25 12:44:40 -05:00
d7c9c181e1 Fixed bug in character class implementation 2025-01-24 19:48:53 -05:00
5a085907cf WIP - fixing character classes 2025-01-24 17:06:19 -05:00
1520edad55 Enforce the rule that character classes must have at least one character; interpret literal closing brackets as regular characters 2025-01-24 15:50:36 -05:00
9d3c228ace Fixed edge cases with character ranges and character classes 2025-01-24 14:57:47 -05:00
2937f2d917 Removed old comment 2025-01-22 20:27:35 -05:00
efab70f9dc Implemented character range detection later in the code, using a metacharacter 2025-01-22 20:26:58 -05:00
649485f01d Removed character range creation from the first part of shuntingYard() (the part that adds concatenation operators), because octal and hex values haven't yet been deciphered at this point in the code 2025-01-22 16:51:00 -05:00
ae09462bd4 Added important note 2025-01-21 22:19:37 -05:00
25cb79f01b Changed the value of EPSILON, so that we can use the NUL character
(which it used to be) in a regex; Also added code to detect escaped
backslashes

Specifically, I replace an escaped backslash with a metacharacter, then
replace it back later on. This prevents problems, like detecting whether
the opening bracket is escaped in '\\[a]'.
2025-01-21 22:12:29 -05:00
099612ae7f Bug fixes, changed the way I parse octal values 2025-01-20 18:04:05 -05:00
3a3333b38a New features, changed character class behavior
I added support for hex values (eg. \x0F), octal values (eg. \012) and
extended hex values (eg. \x{000F2A}). I also expanded the abilities of
character clsses, to include things like escaped characters (eg. [aefp\)])
and character ranges _inside_ inverted character classes (eg. [^\w] which is
functionally equivalent to [\W]).
2025-01-19 21:26:56 -06:00
0b84806fc4 Added 'flags' to the Compile function, instead of maintaining global state to check whether certain features were enabled 2025-01-09 10:33:56 -06:00
c8613c1ba2 Major restructuring - added new type, changed return types for shuntingYard and thompson
I added a new function 'Compile' that calls shuntingYard and thompson. I also added
a new type 'Reg' that this function returns - it represents the starting state and contains
the number of capturing groups in the regex. I also rewrote shuntingYard and thompson
to return errors instead of panicking.
2025-01-06 20:08:24 -06:00
ddbcb309b0 Made shuntingYard return an error instead of panicking, moved it and thompson to compile.go 2025-01-06 12:29:04 -06:00