From 8e9f0b186745afd51579d2a6136a57705efc7574 Mon Sep 17 00:00:00 2001
From: Charlie Stanton <charlie@shtanton.xyz>
Date: Tue, 18 Apr 2023 12:47:55 +0100
Subject: Adds the repeat construct, obsoleting maximise, minimise, try, maybe
 and probably more

The repeat construct repeats a subex a number of times, this number is based on a provided list which is ordered by priority and can be unbounded.
---
 subex/parse.go | 76 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 75 insertions(+), 1 deletion(-)

(limited to 'subex/parse.go')

diff --git a/subex/parse.go b/subex/parse.go
index f2c77bc..24ae082 100644
--- a/subex/parse.go
+++ b/subex/parse.go
@@ -29,6 +29,68 @@ func parseTerminatorAtomLiteral(termType rune, l *RuneReader) walk.Atom {
 	}
 }
 
+func charIsDigit(c rune) bool {
+	return '0' <= c && c <= '9'
+}
+
+// Parse a positive integer, reads digits 0-9 and stops at the first non-digit
+func parseInt(l *RuneReader) (output int) {
+	for {
+		char := l.next()
+		if charIsDigit(char) {
+			output = output * 10 + int(char - '0')
+		} else {
+			break
+		}
+	}
+	l.rewind()
+	return output
+}
+
+// Having just read {, read in and parse the range contents
+func parseRepeatRange(l *RuneReader) (output []ConvexRange) {
+	loop: for {
+		var start, end int
+		char := l.next()
+		l.rewind()
+		if char == '-' {
+			start = -1
+		} else {
+			start = parseInt(l)
+		}
+		switch l.next() {
+			case ',':
+				output = append(output, ConvexRange{start, start})
+				continue loop
+			case '-':
+				char := l.next()
+				if charIsDigit(char) {
+					l.rewind()
+					end = parseInt(l)
+				} else {
+					l.rewind()
+					end = -1
+				}
+			case '}':
+				output = append(output, ConvexRange{start, start})
+				break loop
+			default:
+				panic("Invalid character in repeat specifier")
+		}
+		switch l.next() {
+			case ',':
+				output = append(output, ConvexRange{start, end})
+				continue loop
+			case '}':
+				output = append(output, ConvexRange{start, end})
+				break loop
+			default:
+				panic("Invalid character in repeat specifier")
+		}
+	}
+	return output
+}
+
 func parseReplacement(l *RuneReader) (output []OutputContent) {
 	// TODO escaping
 	loop: for {
@@ -144,7 +206,7 @@ func parseSubex(l *RuneReader, minPower int) SubexAST {
 		case '[':
 			rangeParts := parseRangeSubex(l)
 			lhs = SubexASTRange {rangeParts}
-		case ')', '*', '-', '|', '!', '?', ';':
+		case ')', '*', '-', '|', '!', '?', ';', '{':
 			l.rewind()
 			return nil
 		case '$':
@@ -180,6 +242,11 @@ func parseSubex(l *RuneReader, minPower int) SubexAST {
 		}
 		r := l.next()
 		switch {
+			case r == '{' && minPower <= 8:
+				lhs = SubexASTRepeat{
+					content: lhs,
+					acceptable: parseRepeatRange(l),
+				}
 			case r == '*' && minPower <= 8:
 				lhs = SubexASTMaximise{lhs}
 			case r == '-' && minPower <= 8:
@@ -203,6 +270,13 @@ func parseSubex(l *RuneReader, minPower int) SubexAST {
 					content: lhs,
 					delimiter: rhs,
 				}
+			//case r == '+' && minPower <= 6:
+			//	rhs := parseSubex(l, 7)
+			//	if rhs == nil {
+			//		panic("Missing subex after +")
+			//	}
+			//	// TODO: Implement this. Runs subex on the left, then subex on the right, then sums the outputs of each and outputs that
+			//	lhs = SubexASTAdd{lhs, rhs}
 			default:
 				l.rewind()
 				break loop
-- 
cgit v1.2.3