Patch: ZipInputStream fix

Tom Tromey tromey@cygnus.com
Thu May 11 10:04:00 GMT 2000


I'm checking this in.  This fixes a ZipInputStream bug I found while
reading the code a couple weeks ago.  The `PK\01\02' entries don't
have any data associated with them, so skipping `size' bytes is
definitely wrong.

2000-05-11  Tom Tromey  <tromey@cygnus.com>

	* java/util/zip/ZipInputStream.java (getNextEntry): When reading
	file headers, don't include `size' in the skip call.

Tom

Index: java/util/zip/ZipInputStream.java
===================================================================
RCS file: /cvs/java/libgcj/libjava/java/util/zip/ZipInputStream.java,v
retrieving revision 1.6
diff -u -r1.6 ZipInputStream.java
--- ZipInputStream.java	2000/03/07 19:55:28	1.6
+++ ZipInputStream.java	2000/05/11 16:57:58
@@ -1,4 +1,4 @@
-/* Copyright (C) 1999  Free Software Foundation
+/* Copyright (C) 1999, 2000  Free Software Foundation
 
    This file is part of libgcj.
 
@@ -52,7 +52,13 @@
 	int fname_length = readu2();
 	int extra_length = readu2();
 	int fcomment_length = readu2();
-	in.skip(12+fname_length+extra_length+fcomment_length+size);
+	// `12' is the number of bytes between the comment length
+	// field and the end of the fixed part of the header:
+	// 2 bytes for `disk number start'
+	// 2 bytes for `internal file attributes'
+	// 4 bytes for `external file attributes'
+	// 4 bytes for `relative offset of local header'
+	in.skip(12 + fname_length + extra_length + fcomment_length);
 	if (in.read() != 'P' || in.read() != 'K')
 	  return null;
 	code = in.read();


More information about the Java-patches mailing list