- start: u32
- anchored: tries to match right here
- start_unanchored: u32
- lets the match begin at any later byte, at lower priority
- slot_count: int
- 2 per group, counting the whole match as group 0
- names: [] string
- names[i] is group i's name, "" if unnamed
- anchored: bool
- every match starts at the start of the text (\A)
- has_looks: bool
- has_unicode_word: bool
- \b and friends in Unicode mode, which byte automata can't see
- byte_class: [256] u8
- bytes that the program never tells apart share a class
- class_count: int
- class_byte: [256] u8
- one byte of each class