Tim Ellison wrote:
On 26/Nov/2009 18:44, Tim Ellison wrote:
On 26/Nov/2009 17:00, Oliver Deakin wrote:
<snip>
I'll bet you a beer that it is our friend available() again...
Consider this:
static final String NAME = "javax/swing/text/html/parser/html32.bdtd";
public void test() throws IOException {
JarFile jar = new JarFile("swing.jar");
ZipEntry ze = jar.getEntry(NAME);
InputStream is = jar.getInputStream(ze);
System.out.println("Size = " + ze.getSize());
System.out.println("Available = " + is.available());
jar.close();
}
On Harmony it prints:
Size = 51140
Available = 1
On the RI it prints:
Size = 51140
Available = 51140
I know the arguments that say available() should not be used to judge
the total number of bytes that can be read, but it seems that a number
of applications (including our generated parser?) use it in this way.
I haven't tried changing it yet[1], but I'll take that bet if you want :-)
Regards,
Tim
[1] I'm supposed to be packing for a long weekend away, so I'll be quiet
for the next few days.
Ok, so I may owe you a beer :-)
Sounds good!
A quick trial of implementing available() for the inflater stream fixes
this test case (above) but not the original problem.
Yep, I see the same thing. Although (as Jesse mentioned in the JIRA) we
call available() in DTD.java and the results are significantly different
for M11 and M12, as exposed by your test case above, implementing
available() still does not fix the swing test case.
Given there are these differences we see emerging, I wonder if we should
revert the offending code and retry again after M12?
It would be tempting to do this so we can get M12 out the door and then
fix this issue for M13.
FYI here is my hacked available() impl, which needs some testing before
it is good to go in..
Index: src/main/java/java/util/zip/ZipFile.java
===================================================================
--- src/main/java/java/util/zip/ZipFile.java (revision 884543)
+++ src/main/java/java/util/zip/ZipFile.java (working copy)
@@ -247,7 +247,7 @@
rafstrm.skip(entry.nameLen + localExtraLenOrWhatever);
rafstrm.mLength = rafstrm.mOffset + entry.compressedSize;
if (entry.compressionMethod == ZipEntry.DEFLATED) {
- return new InflaterInputStream(rafstrm, new
Inflater(true));
+ return new ZipInflaterInputStream(rafstrm, new
Inflater(true), 1024, entry);
} else {
return rafstrm;
}
@@ -415,4 +415,27 @@
return n;
}
}
+
+ static class ZipInflaterInputStream extends InflaterInputStream {
+
+ ZipEntry entry;
+ long bytesRead = 0;
+
+ public ZipInflaterInputStream(InputStream is, Inflater inf, int
bsize, ZipEntry entry) {
+ super(is, inf, bsize);
+ this.entry = entry;
+ }
+
+ public int read(byte[] buffer, int off, int nbytes) throws
IOException {
+ int i = super.read(buffer, off, nbytes);
+ if (i != -1) {
+ bytesRead += i;
+ }
+ return i;
+ }
+
+ public int available() throws IOException {
+ return super.available() == 0 ? 0 : (int) (entry.getSize()
- bytesRead);
+ }
+ }
}
--
Oliver Deakin
Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU
