3148-backport-2395-to-Pharo-70--Non-ASCII-class-and-author-names-break-SourceFileArraygetPreambleFromat-

src/System-Sources/SourceFile.class.st

@@ -61,6 +61,34 @@ SourceFile >> fullName [
 	^ path asString
 ]
 
+{ #category : #accessing }
+SourceFile >> getPreambleAt: startingPosition [
+	"Search backwards from byte startingPosition in my stream for a method preamble and return it.
+	A method preamble looks like: MyClass methodsFor: 'test' stamp: 'author 1/27/2019 12:27'
+	but with exclamation marks ($!) around it (the contents excluding the $!'s is returned).
+	startingPosition should be set one position before the closing $!"
+
+	| characterReadStream binaryStream encoder position |
+	"I hold either a ZnCharacterReadStream or a ZnCharacterReadWriteStream (see #tryOpenReadOnly:)
+	Use #isReadOnly and #readOnlyCopy to access the ZnCharacterReadStream in both cases"
+	characterReadStream := self isReadOnly ifTrue: [ stream ] ifFalse: [ stream readOnlyCopy ].
+	"Access the binary read stream wrapped by the character read stream"
+	binaryStream := characterReadStream wrappedStream.
+	"Access the encoder held by the character read stream"
+	encoder := characterReadStream encoder.
+	"Search backwards for the previous occurrence of $!
+	Although the underlying encoding is UTF-8 we can still operate/move at the byte level
+	since $! code 33 cannot occur in code points encoded using 2, 3 or 4 bytes"
+	position := startingPosition.
+	[ position >= 0 
+		and: [ 
+			binaryStream position: position. 
+			binaryStream next ~= 33 "$!" ] ] 
+		whileTrue: [ position := position - 1 ].
+	"Now that we found the byte range, extract and decode it"
+	^ encoder decodeBytes: (binaryStream next: startingPosition - position)
+]
+
 { #category : #testing }
 SourceFile >> isOpen [
 

src/System-Sources/SourceFileArray.class.st

@@ -329,17 +329,7 @@ SourceFileArray >> forceChangesToDisk [
 
 { #category : #'public - string reading' }
 SourceFileArray >> getPreambleFrom: aFileStream at: position [
-	"To read preamble of method we need read back characters until $!.
-	But given aFileStream can have UTF8 encoding which make it a bit tricky to scan stream in reverse order.
-	First we need to move back in stream using byte reading without encoding. It performed by basicNext.
-	Next we still could not read exact number of characters because (position-startIndex) is not characters number in case of UTF8.
-	So at the end we use another loop to read encoded characters step by step until stream will be at original position"
-
-	| startIndex |
-	startIndex := position.
-	[ startIndex >= 0 and: [ aFileStream position: startIndex. aFileStream next ~~ $! ] ] whileTrue: [ startIndex := startIndex - 1 ].
-
-	^ String streamContents: [ :result | result nextPutAll: (aFileStream next: position - aFileStream position + 1) ]
+	^ aFileStream getPreambleAt: position 
 ]
 
 { #category : #initialization }