Add handling for logging and gtest comparison macros.

Committed: https://crrev.com/5efc4e117639ad7f3d1093f463bab625802eee0f
Cr-Commit-Position: refs/heads/master@{#291615}

[dcheng, 2014-08-20 06:58:00]

After a surprisingly frustrating series of experiments with clang-query, I have
a CL that should hopefully handle the macros that confuse the tool.

[dcheng, 2014-08-20 07:02:30]

https://codereview.chromium.org/488933002/diff/20001/tools/clang/rewrite_scop...
File tools/clang/rewrite_scoped_refptr/RewriteScopedRefptr.cpp (right):

https://codereview.chromium.org/488933002/diff/20001/tools/clang/rewrite_scop...
tools/clang/rewrite_scoped_refptr/RewriteScopedRefptr.cpp:71:
assert(range.isValid() && "Invalid range!");
This is apparently LLVM style.

https://codereview.chromium.org/488933002/diff/20001/tools/clang/rewrite_scop...
tools/clang/rewrite_scoped_refptr/RewriteScopedRefptr.cpp:137: Replacement
RewriteRawPtrToScopedRefptr(const MatchFinder::MatchResult& result,
This is just a move of the original code, re-adding back some of the early
returns {that existed before it was turned into a function} in the form of
asserts.

[dcheng, 2014-08-22 19:35:49]

ping!

[Ryan Sleevi, 2014-08-23 03:26:36]

lgtm

https://codereview.chromium.org/488933002/diff/40001/tools/clang/scripts/test...
File tools/clang/scripts/test_tool.py (right):

https://codereview.chromium.org/488933002/diff/40001/tools/clang/scripts/test...
tools/clang/scripts/test_tool.py:54:
os.path.realpath(os.path.join(tools_clang_directory, '../..')))
Ooof. I didn't realize we'd done this.

I had tried to keep the original tool fully hermetic. I can understand the need
for gtest, but it seems unfortunate to mix tool and Chromium code.

[dcheng, 2014-08-24 02:41:38]

On 2014/08/23 at 03:26:36, rsleevi wrote:
> lgtm
> 
>
https://codereview.chromium.org/488933002/diff/40001/tools/clang/scripts/test...
> File tools/clang/scripts/test_tool.py (right):
> 
>
https://codereview.chromium.org/488933002/diff/40001/tools/clang/scripts/test...
> tools/clang/scripts/test_tool.py:54:
os.path.realpath(os.path.join(tools_clang_directory, '../..')))
> Ooof. I didn't realize we'd done this.
> 
> I had tried to keep the original tool fully hermetic. I can understand the
need for gtest, but it seems unfortunate to mix tool and Chromium code.

It is unfortunate, but I think this was slightly less bad:
- If I keep it hermetic, I have to duplicate the gtest/logging cruft. And it
means if the logging/gtest macros change for some reason, the tool will probably
behave unexpectedly.
- If I don't keep it hermetic, the tool grows weird dependencies on Chromium.
And the tests will probably break for platforms I finish converting. Since I
don't expect to run the tool on a fully converted platform, I think this is
acceptable.
- Because I started depending on base/logging.h, I decided to use the
scoped_refptr from base too just to be safe.

[dcheng, 2014-08-24 02:43:03]

Committed patchset #3 to pending queue manually as 4ee6a1d (presubmit
successful).

[commit-bot: I haz the power, 2014-09-10 02:32:18]

Patchset 3 (id:??) landed as
https://crrev.com/5efc4e117639ad7f3d1093f463bab625802eee0f
Cr-Commit-Position: refs/heads/master@{#291615}