1 files changed, 175 insertions, 0 deletions
diff --git a/subex/parse.go b/subex/parse.go
new file mode 100644
index 0000000..af575eb
--- /dev/null
+++ b/subex/parse.go
@@ -0,0 +1,175 @@
+package subex
+
+func parseReplacement(l *RuneReader) (output []TransducerOutput) {
+	loop: for {
+		r := l.next()
+		switch r {
+			case eof:
+				panic("Missing closing \"")
+			case '"':
+				break loop
+			case '$':
+				slot := l.next()
+				if slot == eof {
+					panic("Missing slot character")
+				}
+				output = append(output, TransducerReplacementLoad(slot))
+			default:
+				output = append(output, TransducerReplacementRune(r))
+		}
+	}
+	return output
+}
+
+func parseRangeSubex(l *RuneReader) map[rune]rune {
+	parts := make(map[rune]rune)
+	var froms []rune
+	var hasTo bool
+	for {
+		fromsStart := l.next()
+		if fromsStart == ']' {
+			hasTo = false
+			break
+		} else if fromsStart == '=' {
+			hasTo = true
+			break
+		}
+		var fromsEnd rune
+		if l.accept("-") {
+			fromsEnd = l.next()
+			if fromsEnd == ']' || fromsEnd == '=' {
+				l.rewind()
+				fromsEnd = fromsStart
+			}
+		} else {
+			fromsEnd = fromsStart
+		}
+		for i := fromsStart; i <= fromsEnd; i += 1 {
+			froms = append(froms, i)
+		}
+	}
+	if len(froms) == 0 {
+		panic("Missing from part of range expression")
+	}
+
+	var tos []rune
+	if hasTo {
+		for {
+			tosStart := l.next()
+			if tosStart == ']' {
+				break
+			}
+			var tosEnd rune
+			if l.accept("-") {
+				tosEnd = l.next()
+				if tosEnd == ']' {
+					l.rewind()
+					tosEnd = tosStart
+				}
+			} else {
+				tosEnd = tosStart
+			}
+			for i := tosStart; i <= tosEnd; i += 1 {
+				tos = append(tos, i)
+			}
+		}
+	} else {
+		tos = froms
+	}
+	if len(tos) == 0 {
+		panic("Missing to part of range expression")
+	}
+	
+	for i, from := range froms {
+		parts[from] = tos[i % len(tos)]
+	}
+	return parts
+}
+
+func parseSubex(l *RuneReader, minPower int) SubexAST {
+	var lhs SubexAST
+	r := l.next()
+	switch r {
+		case eof:
+			return nil
+		case '(':
+			lhs = parseSubex(l, 0)
+			if !l.accept(")") {
+				panic("Missing matching )")
+			}
+		case '[':
+			rangeParts := parseRangeSubex(l)
+			lhs = SubexASTRange {rangeParts}
+		case ')', '*', '-', '|', '!', '?', ';':
+			l.rewind()
+			return nil
+		case '$':
+			slot := l.next()
+			if slot == eof {
+				panic("Missing slot character")
+			}
+			match := parseSubex(l, 100)
+			if match == nil {
+				panic("Missing regex for store")
+			}
+			lhs = SubexASTStore{
+				match: match,
+				slot: slot,
+			}
+		case '"':
+			replacement := parseReplacement(l)
+			lhs = SubexASTOutput{replacement}
+		case '.':
+			lhs = SubexASTCopyAny{}
+		default:
+			lhs = SubexASTCopyRune(r)
+	}
+	loop: for {
+		if minPower <= 0 {
+			next := parseSubex(l, 1)
+			if next != nil {
+				lhs = SubexASTConcat{lhs, next}
+				continue loop
+			}
+		}
+		r := l.next()
+		switch {
+			case r == '*' && minPower <= 8:
+				lhs = SubexASTMaximise{lhs}
+			case r == '-' && minPower <= 8:
+				lhs = SubexASTMinimise{lhs}
+			case r == '!' && minPower <= 8:
+				lhs = SubexASTTry{lhs}
+			case r == '?' && minPower <= 8:
+				lhs = SubexASTMaybe{lhs}
+			case r == '|' && minPower <= 4:
+				rhs := parseSubex(l, 5)
+				if rhs == nil {
+					panic("Missing subex after |")
+				}
+				lhs = SubexASTOr{lhs, rhs}
+			case r == ';' && minPower <= 2:
+				rhs := parseSubex(l, 3)
+				if rhs == nil {
+					panic("Missing subex after ;")
+				}
+				lhs = SubexASTJoin{
+					content: lhs,
+					delimiter: rhs,
+				}
+			default:
+				l.rewind()
+				break loop
+		}
+	}
+	return lhs
+}
+
+func Parse(input string) SubexAST {
+	l := RuneReader {
+		input: input,
+		pos: 0,
+		width: 0,
+	}
+	return parseSubex(&l, 0)
+}