[PreSmalltalks] Implement LH "Unsafe Symbol" encoding

src/BaselineOfMachineArithmetic/BaselineOfMachineArithmetic.class.st

@@ -31,11 +31,13 @@ BaselineOfMachineArithmetic >> baseline: spec [
 				package:#'PreSmalltalks-Applicative';
 				package:#'PreSmalltalks-Substitutions';
 				package:#'PreSmalltalks-ZEncoding';
+				package:#'PreSmalltalks-FxEncoding';
 				package: #'PreSmalltalks' with:
 					[spec requires: 'PreSmalltalks-Pharo';
 					      requires: 'PreSmalltalks-Applicative';
 					      requires: 'PreSmalltalks-Substitutions';
-					      requires: 'PreSmalltalks-ZEncoding'
+					      requires: 'PreSmalltalks-ZEncoding';
+					      requires: 'PreSmalltalks-FxEncoding'
 					];
 				package: #'PreSmalltalks-Parser' with:
 					[spec requires: 'PreSmalltalks';

src/PreSmalltalks-FxEncoding/Character.extension.st

@@ -0,0 +1,18 @@
+Extension { #name : #Character }
+
+{ #category : #'*PreSmalltalks-FxEncoding' }
+Character >> isUnsafeChar [
+"
+isUnsafeChar :: Char -> Bool
+Cf. Names.hs
+"
+	^(Character okSymChars includes: self) not
+]
+
+{ #category : #'*PreSmalltalks-FxEncoding' }
+Character class >> okSymChars [
+	^($a to: $z),
+	 ($A to: $Z),
+	 ($0 to: $9),
+	 {$_}
+]

src/PreSmalltalks-FxEncoding/String.extension.st

@@ -0,0 +1,43 @@
+Extension { #name : #String }
+
+{ #category : #'*PreSmalltalks-FxEncoding' }
+String >> decodeUnsafe [
+	| in out |
+	in := ReadStream on: self.
+	out := WriteStream on: String new.
+	[ in atEnd ] whileFalse: [ | c |
+			c := in next.
+			c = $ß
+				ifTrue: [ | codePoint |
+					codePoint := Integer readFrom: (in upTo: $ß).
+					out nextPut: (Character codePoint: codePoint)
+				] ifFalse: [
+					out nextPut: c
+				]
+		].
+	^out contents
+	"
+	'vß35ß3ß35ßß35ß1' decodeUnsafe  → 'v#3##1'
+	"
+]
+
+{ #category : #'*PreSmalltalks-FxEncoding' }
+String >> encodeUnsafe [
+"
+encodeUnsafe :: T.Text -> T.Text
+Cf. Names.hs
+In Smalltalk, we want to satisfy stricter requirements:
+we want our symbols to be valid Z3 symbols (even after passing
+through FFI) AND be valid Smalltalk names.
+Therefore '$' is not a safe character; but 'ß' is.
+"
+	| c cs head tail |
+	self isEmpty ifTrue: [ ^ self ].
+	c := self first.
+	cs := self allButFirst.
+	head := c isUnsafeChar
+		ifTrue: [ 'ß' , c codePoint printString , 'ß' ]
+		ifFalse: [ String with: c ].
+	tail := cs encodeUnsafe.
+	^ head , tail
+]

src/PreSmalltalks-FxEncoding/package.st

@@ -0,0 +1 @@
+Package { #name : #'PreSmalltalks-FxEncoding' }