Revert 52799 - Add support for parsing certificate formats other than raw, DE...

Revert 52799 - Add support for parsing certificate formats other than raw, DER-encoded certificates - specifically formats that represent collections of certificates. The certificate format can now be specified as an explicit format, or as a bit-mask of formats that are acceptable/expected, with the first parsable format winning.

This is one half of a commit to address BUG #37142, with the second half involving connecting this through the X509UserCertHandler and the actual UI.

R=wtc
BUG=37142
TEST=X509CertificateParseTest* and PEMTokenizerTest.*

Review URL: http://codereview.chromium.org/2819018

TBR=rsleevi@chromium.org

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=52801

[wtc, 2010-07-17 13:28:58]

When you revert a CL, please add a comment to note why
you had to revert it.

The CL you reverted broke the Mac builds.  I found some
compiler warnings in x509_certificate_unittest.cc that
were treated as errors.  I don't know what those warnings
means.  I hope you can figure out how to fix them.

http://codereview.chromium.org/2812064/diff/1/9
File net/base/x509_certificate_unittest.cc (left):

http://codereview.chromium.org/2812064/diff/1/9#oldcode669
net/base/x509_certificate_unittest.cc:669: for (size_t j = 0; j < 20; ++j)
The constant 20 should be replaced by
sizeof(actual_fingerprint.data).

[rsleevi-old, 2010-07-17 18:22:58]

On 2010/07/17 13:28:58, wtc wrote:
> When you revert a CL, please add a comment to note why
> you had to revert it.
> 
> The CL you reverted broke the Mac builds.  I found some
> compiler warnings in x509_certificate_unittest.cc that
> were treated as errors.  I don't know what those warnings
> means.  I hope you can figure out how to fix them.
> 
> http://codereview.chromium.org/2812064/diff/1/9
> File net/base/x509_certificate_unittest.cc (left):
> 
> http://codereview.chromium.org/2812064/diff/1/9#oldcode669
> net/base/x509_certificate_unittest.cc:669: for (size_t j = 0; j < 20; ++j)
> The constant 20 should be replaced by
> sizeof(actual_fingerprint.data).

Yeah, this was my first time using drover, and it didn't like my cygwin
environment that I was running from (Specifically, it wasn't able to find fi,
which would have allowed me to edit the message). I was given the option of futz
with it, or just revert.

I've fixed the issue in the original CL. for whatever reason, the OS X 10.5 gcc
4.2.1, unlike all the other boxes/compilers, didn't like some of the data being
in an anonymous namespace. The simple fix was to remove the anonymous namespace
qualifier in the x509_certificate_unittests. This doesn't have any functional
effect, even to linking, since all of the implementation is contained within
that file and only used by the unit tests.

This non-use of anon was consistant with the other places in code using
instantiated tests, so it seems correct. I was just waiting to reland until I
had a chance to babysit the tree again.

Trybots compile it fine (though they fail the tests, because the trybots don't
like binary files), so it's all good.