Fixed inverted Perl classes inside character classes eg. '[\D]'; added new flag for 'single-line' mode, where dot metacharacter matches newlines

master
Aadhavan Srinivasan 5 days ago
parent 61a128e27a
commit 4c3c7e21c5

@ -24,9 +24,10 @@ const CONCAT rune = '~'
type ReFlag int
const (
RE_NO_FLAGS ReFlag = iota
RE_CASE_INSENSITIVE
RE_MULTILINE
RE_NO_FLAGS ReFlag = iota
RE_CASE_INSENSITIVE // Case insensitive matching
RE_MULTILINE // '^' and '$' assert at start and end of _line_, rather than start and end of input string
RE_SINGLE_LINE // Dot metacharacter matches newline characters.
)
func isOperator(c rune) bool {
@ -78,6 +79,8 @@ func getPOSIXClass(str []rune) (bool, string) {
var caseInsensitive bool
// Stores whether the multiline flag has been enabled.
// In multi-line mode, '^' and '$' assert position at the start and
// end of a _line_ rather than the entire input.
var multilineMode bool
/*
@ -93,10 +96,10 @@ func shuntingYard(re string, flags ...ReFlag) ([]postfixNode, error) {
caseInsensitive = false
multilineMode = false
// In Multiline mode, the newline character is considered a
// 'dot' character ie. the dot metacharacter matches a newline as well.
if slices.Contains(flags, RE_MULTILINE) {
multilineMode = true
}
if slices.Contains(flags, RE_SINGLE_LINE) {
notDotChars = []rune{}
} else {
notDotChars = []rune{'\n'}
@ -965,6 +968,12 @@ func thompson(re []postfixNode) (Reg, error) {
})...)
}
s.content = rune2Contents(nodeContents)
if len(node.except) > 0 {
s.allChars = true
s.except = slices.Concat(Map(node.except, func(n postfixNode) []rune {
return n.contents
})...)
}
return &s
})
// Reduce the list of states down to a single state by alternating them

Loading…
Cancel
Save