subex

commit b2d1d965dee8cc2c1e063067d53a3c8e28a46d6c
parent ce2db2bc333ed938ec93d5ad0838f8cb720c4865
Author: Charlie Stanton <charlie@shtanton.xyz>
Date:   Sat, 24 Dec 2022 10:04:42 +0000

Adds the character range mapping syntax

Ranges of characters can be mapped with []

For example, capitalisation of a letter: [a-z=A-Z]
Caesar cipher shift of 1: [a-zA-Z=b-zaB-ZA]

Diffstat:
M
main/parse.go
 | 
68
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
M
main/subexast.go
 | 
10
++++++++++
M
main/subexstate.go
 | 
20
++++++++++++++++++++

3 files changed, 98 insertions(+), 0 deletions(-)
diff --git a/main/parse.go b/main/parse.go
@@ -21,6 +21,71 @@ func parseReplacement(l *RuneReader) (output []TransducerOutput) {
 	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()
@@ -32,6 +97,9 @@ func parseSubex(l *RuneReader, minPower int) SubexAST {
 			if !l.accept(")") {
 				panic("Missing matching )")
 			}
+		case '[':
+			rangeParts := parseRangeSubex(l)
+			lhs = SubexASTRange {rangeParts}
 		case ')', '*', '-', '|', '!', '?', ';':
 			l.rewind()
 			return nil
diff --git a/main/subexast.go b/main/subexast.go
@@ -151,3 +151,13 @@ func (ast SubexASTJoin) compileWith(next SubexState) SubexState {
 		next,
 	}
 }
+
+type SubexASTRange struct {
+	parts map[rune]rune
+}
+func (ast SubexASTRange) compileWith(next SubexState) SubexState {
+	return &SubexRangeState {
+		parts: ast.parts,
+		next: next,
+	}
+}
diff --git a/main/subexstate.go b/main/subexstate.go
@@ -123,3 +123,23 @@ func (state SubexCopyAnyState) eat(store Store, char rune) []SubexBranch {
 func (state SubexCopyAnyState) accepting(store Store) []string {
 	return nil
 }
+
+type SubexRangeState struct {
+	parts map[rune]rune
+	next SubexState
+}
+func (state SubexRangeState) eat(store Store, char rune) []SubexBranch {
+	out, exists := state.parts[char]
+	if !exists {
+		return nil
+	} else {
+		return []SubexBranch{{
+			state: state.next,
+			output: string(out),
+			store: store,
+		}}
+	}
+}
+func (state SubexRangeState) accepting(store Store) []string {
+	return nil
+}