OpenSystemsDevelopment / qpdf

Try a  limit of 50MB. For very large limits processing time before
damage is encountered may exceed oss-fuzz limits.
Add further test cases.

Reject non-dictionary Page and Pages objects.

Also add additional qpdf_fuzzer test cases.

Add test case for oss-fuzz 15471 and 69977a

Fixes oss-fuzz 70055

Add extra fuzz test case and amend memory limit for Pl_DCT.

Also, add diagnostic messages in qpdf_fuzzer and additional fuzz test case.

In QPDF::read_xrefEntry add buffer overflow test for first eol character.
Overlong f1 or f2 entries consisting only of zeros could cause a buffer
overflow.

Add fuzz testcase 69913.

Also add new fuzz test case.

Code failed to allow for QPDF::getCompressibleObjSet deleting objects
from the object cache in case of multiple entries for the same object id.

Add fuzz test case 68668.

Add fuzz case 68377.

...as well as some cases generated in CI from earlier attempts at
fixing this.

Fix two errors introduced in #1110 and #1112. Since
#1110, encountering the invalid indirect reference #1110
-2147483648 n R produces an integer underflow which, if
 undetected, immediately trigger a logic error. Since
 #1112, object -1 0 R may be incorrectly identified as
 an earlier generation of itself and deleted,
 invalidating a live iterator.

It is possible to reproduce the failure with this file following the
instructions with oss-fuzz, though it does not cause a failure in CI.

The failure was introduced in
18c1ffe0df335a46cddbeb96e2cb939d850df9fa.

Files are copied into the build area rather than left in the source
tree, and the test suite looks for them there. Also remove special
case around counting files in the qpdf corpus.

Since the introduction of fuzz testing, there has never been a problem
found because of a failure of a file in the fuzzer seed corpus. As the
fuzzer has found problems, they have been added to the test suite, and
that should be adequate to exercise the fuzzers in the tesing
environment as well as providing adequate regression testing.

Removing these original files shaves many minutes off the builds in CI.

When making resources indirect in from_dr, the code was using the
wrong owning QPDF, forgetting that from_dr had already been copied
using CopyForeignObject.

There isn't really an issue with these files causing a real problem,
but malware and virus checkers trip on them, and the value to leaving
them in the test suite is too low to be worth the hassle.

It's detected in QPDFWriter instead of at parse time because I can't
figure out how to construct a test case in a reasonable time. This
commit moves the fuzz file into the regular test suite for a QTC
coverage case.

This makes it faster to iterate on the other ones.

* Create the seed corpus in the build directory
* Don't assume all fuzzers share an options file

Temporarily skip fuzz tests on Windows. There are Windows-specific
failures to address later.

qutil.cc uses strerror to print some exceptions and adds a newline
afterwards, but strerror in Windows already adds one newline at the
end of the message and the additional one from std::endl breaks the
output vs. the expected one.