4545import static com .oracle .graal .python .runtime .exception .PythonErrorType .UnicodeDecodeError ;
4646import static com .oracle .graal .python .runtime .exception .PythonErrorType .UnicodeEncodeError ;
4747
48- import java .nio .BufferUnderflowException ;
4948import java .nio .ByteBuffer ;
5049import java .nio .CharBuffer ;
5150import java .nio .charset .CharacterCodingException ;
@@ -258,6 +257,7 @@ private int getLength(PBytes b) {
258257 }
259258 }
260259
260+ // Encoder for raw_unicode_escape
261261 @ Builtin (name = "__truffle_raw_encode" , minNumOfPositionalArgs = 1 , parameterNames = {"str" , "errors" })
262262 @ GenerateNodeFactory
263263 public abstract static class RawEncodeNode extends EncodeBaseNode {
@@ -425,14 +425,15 @@ private boolean castToBoolean(VirtualFrame frame, Object object) {
425425 }
426426 }
427427
428+ // Decoder for raw_escape_unicode
428429 @ Builtin (name = "__truffle_raw_decode" , minNumOfPositionalArgs = 1 , parameterNames = {"bytes" , "errors" })
429430 @ GenerateNodeFactory
430431 abstract static class RawDecodeNode extends EncodeBaseNode {
431432 @ Child private GetInternalByteArrayNode toByteArrayNode ;
432433
433434 @ Specialization
434435 Object decode (PBytesLike bytes , @ SuppressWarnings ("unused" ) PNone errors ) {
435- String string = decodeBytes (getBytesBuffer (bytes ), "strict" );
436+ String string = decodeBytes (getBytes (bytes ), "strict" );
436437 return factory ().createTuple (new Object []{string , string .length ()});
437438 }
438439
@@ -446,60 +447,64 @@ Object decode(PBytesLike bytes, Object errors,
446447 CompilerDirectives .transferToInterpreterAndInvalidate ();
447448 throw new IllegalStateException ("should not be reached" );
448449 }
449- String string = decodeBytes (getBytesBuffer (bytes ), profiledErrors );
450+ String string = decodeBytes (getBytes (bytes ), profiledErrors );
450451 return factory ().createTuple (new Object []{string , string .length ()});
451452 }
452453
453- private ByteBuffer getBytesBuffer (PBytesLike bytesLike ) {
454+ private byte [] getBytes (PBytesLike bytesLike ) {
454455 if (toByteArrayNode == null ) {
455456 CompilerDirectives .transferToInterpreterAndInvalidate ();
456457 toByteArrayNode = insert (GetInternalByteArrayNodeGen .create ());
457458 }
458- byte [] barr = toByteArrayNode .execute (bytesLike .getSequenceStorage ());
459- return ByteBuffer .wrap (barr , 0 , barr .length );
459+ return toByteArrayNode .execute (bytesLike .getSequenceStorage ());
460460 }
461461
462462 @ TruffleBoundary
463- String decodeBytes (ByteBuffer bytes , String errors ) {
463+ String decodeBytes (byte [] bytes , String errors ) {
464464 CodingErrorAction errorAction = convertCodingErrorAction (errors );
465465 try {
466- ByteBuffer buf = ByteBuffer .allocate (bytes .remaining () * Integer .BYTES );
467- byte [] hexString = new byte [ 8 ] ;
468- while (bytes .hasRemaining () ) {
469- int val ;
470- byte b = bytes .get ();
471- if ( b == ( byte ) '\\' ) {
472- byte b1 = bytes . get () ;
466+ ByteBuffer buf = ByteBuffer .allocate (bytes .length * Integer .BYTES );
467+ int i = 0 ;
468+ while (i < bytes .length ) {
469+ byte b = bytes [ i ] ;
470+ if ( b == ( byte ) '\\' && i + 1 < bytes .length ) {
471+ byte b1 = bytes [ i + 1 ];
472+ int numIndex = i + 2 ;
473473 if (b1 == (byte ) 'u' ) {
474- bytes .get (hexString , 0 , 4 );
475- val = Integer .parseInt (new String (hexString , 0 , 4 ), 16 );
474+ final int count = 4 ;
475+ if (numIndex + count > bytes .length ) {
476+ throw raise (UnicodeDecodeError );
477+ }
478+ buf .putInt (Integer .parseInt (new String (bytes , numIndex , count ), 16 ));
479+ i = numIndex + count ;
480+ continue ;
476481 } else if (b1 == (byte ) 'U' ) {
477- bytes .get (hexString , 0 , 8 );
478- val = Integer .parseInt (new String (hexString , 0 , 8 ), 16 );
479- } else {
480- throw new CharacterCodingException ();
482+ final int count = 8 ;
483+ if (numIndex + count > bytes .length ) {
484+ throw raise (UnicodeDecodeError );
485+ }
486+ buf .putInt (Integer .parseInt (new String (bytes , numIndex , count ), 16 ));
487+ i = numIndex + count ;
488+ continue ;
481489 }
482- } else {
483- // Bytes that are not an escape sequence are latin-1, which maps to unicode
484- // codepoints directly
485- val = b & 0xFF ;
486490 }
487- buf .putInt (val );
491+ // Bytes that are not an escape sequence are latin-1, which maps to unicode
492+ // codepoints directly
493+ buf .putInt (b & 0xFF );
494+ i ++;
488495 }
489496 buf .flip ();
490497 CharBuffer decoded = UTF32 .newDecoder ().onMalformedInput (errorAction ).onUnmappableCharacter (errorAction ).decode (buf );
491498 return String .valueOf (decoded );
492- } catch (CharacterCodingException | NumberFormatException | BufferUnderflowException e ) {
499+ } catch (CharacterCodingException | NumberFormatException e ) {
493500 throw raise (UnicodeDecodeError , e );
494501 }
495502 }
496503 }
497504
498- // _codecs.lookup(name)
499505 @ Builtin (name = "__truffle_lookup" , minNumOfPositionalArgs = 1 )
500506 @ GenerateNodeFactory
501507 abstract static class CodecsLookupNode extends PythonBuiltinNode {
502- // This is replaced in the core _codecs.py with the full functionality
503508 @ Specialization
504509 Object lookup (String encoding ) {
505510 if (CharsetMapping .getCharset (encoding ) != null ) {
@@ -510,7 +515,6 @@ Object lookup(String encoding) {
510515 }
511516 }
512517
513- // _codecs.lookup(name)
514518 @ Builtin (name = "charmap_build" , minNumOfPositionalArgs = 1 )
515519 @ GenerateNodeFactory
516520 abstract static class CharmapBuildNode extends PythonBuiltinNode {
0 commit comments