[ASan] fix symbolizer script to run llvm-symbolizer

[ramosian.glider, 2012-09-18 14:33:01]

Looks mostly good.
Can you please run pylint on this script before committing?

https://codereview.appspot.com/6536044/diff/2001/scripts/asan_symbolize.py
File scripts/asan_symbolize.py (right):

https://codereview.appspot.com/6536044/diff/2001/scripts/asan_symbolize.py#ne...
scripts/asan_symbolize.py:162: def symbolize(self, addr, binary, offset):
Please add a comment about what this function does. It's getting hard to
understand.

https://codereview.appspot.com/6536044/diff/2001/scripts/asan_symbolize.py#ne...
scripts/asan_symbolize.py:191: if symbolizer is not None:
Looks like "is not None" is a valid idiom, but I'd better avoid it.

https://codereview.appspot.com/6536044/diff/2001/scripts/asan_symbolize.py#ne...
scripts/asan_symbolize.py:290: # (fall back to next symbolizer if the current
fails).
s/current/previous one/

https://codereview.appspot.com/6536044/diff/2001/scripts/asan_symbolize.py#ne...
scripts/asan_symbolize.py:300: # System symbolizer must produce some result.
"The system"

https://codereview.appspot.com/6536044/diff/2001/scripts/asan_symbolize.py#ne...
scripts/asan_symbolize.py:312: match = re.match('^( *#([0-9]+) *)(0x[0-9a-f]+)
*\((.*)\+(0x[0-9a-f]+)\)',
Please make the regexp a constant with a meaningful name so that there's less
garbage in main()

https://codereview.appspot.com/6536044/diff/2001/scripts/asan_symbolize.py#ne...
scripts/asan_symbolize.py:319: prefix = match.group(1)
I wonder if we could initialize all the vars at once to make the code more
clear:

prefix, frameno_str, addr, binary, offset = match.groups()

[samsonov, 2012-09-19 06:30:58]

I've run pylint and it printed quite a number of warnings around the code, so
probably I don't have the config you use. Anyway, I'm ok with addressing them,
but probably it should be a separate commit.

https://codereview.appspot.com/6536044/diff/2001/scripts/asan_symbolize.py
File scripts/asan_symbolize.py (right):

https://codereview.appspot.com/6536044/diff/2001/scripts/asan_symbolize.py#ne...
scripts/asan_symbolize.py:162: def symbolize(self, addr, binary, offset):
On 2012/09/18 14:33:01, ramosian.glider wrote:
> Please add a comment about what this function does. It's getting hard to
> understand.

Done.

https://codereview.appspot.com/6536044/diff/2001/scripts/asan_symbolize.py#ne...
scripts/asan_symbolize.py:191: if symbolizer is not None:
On 2012/09/18 14:33:01, ramosian.glider wrote:
> Looks like "is not None" is a valid idiom, but I'd better avoid it.

Done.

https://codereview.appspot.com/6536044/diff/2001/scripts/asan_symbolize.py#ne...
scripts/asan_symbolize.py:290: # (fall back to next symbolizer if the current
fails).
On 2012/09/18 14:33:01, ramosian.glider wrote:
> s/current/previous one/

Done.

https://codereview.appspot.com/6536044/diff/2001/scripts/asan_symbolize.py#ne...
scripts/asan_symbolize.py:300: # System symbolizer must produce some result.
On 2012/09/18 14:33:01, ramosian.glider wrote:
> "The system"

Done.

https://codereview.appspot.com/6536044/diff/2001/scripts/asan_symbolize.py#ne...
scripts/asan_symbolize.py:312: match = re.match('^( *#([0-9]+) *)(0x[0-9a-f]+)
*\((.*)\+(0x[0-9a-f]+)\)',
On 2012/09/18 14:33:01, ramosian.glider wrote:
> Please make the regexp a constant with a meaningful name so that there's less
> garbage in main()

Done.

https://codereview.appspot.com/6536044/diff/2001/scripts/asan_symbolize.py#ne...
scripts/asan_symbolize.py:319: prefix = match.group(1)
On 2012/09/18 14:33:01, ramosian.glider wrote:
> I wonder if we could initialize all the vars at once to make the code more
> clear:
> 
> prefix, frameno_str, addr, binary, offset = match.groups()

Done.

[ramosian.glider, 2012-09-19 07:39:24]

LGTM
We'll need to clean up the gpylint warnings in a separate CL

[samsonov, 2012-09-19 08:51:49]

On 2012/09/19 07:39:24, ramosian.glider wrote:
> LGTM
> We'll need to clean up the gpylint warnings in a separate CL

r164209