Fix some problems detected by coverity.

Fix some problems detected by coverity.

- Uninitialized class member in GSCanonicalEntry and SkPDFDocument.
- Incorrect sign extension in SkPDFFont.
- Dead code in SkPDFUtils.

CID=16262,16272,16273,16275

Committed: http://code.google.com/p/skia/source/detail?r=1668

[Chris Guillory, 2011-06-21 21:01:11]

LGTM. One nit.

http://codereview.appspot.com/4659041/diff/1/src/pdf/SkPDFFont.cpp
File src/pdf/SkPDFFont.cpp (right):

http://codereview.appspot.com/4659041/diff/1/src/pdf/SkPDFFont.cpp#newcode50
src/pdf/SkPDFFont.cpp:50: *size = (size_t)buf[2] | ((size_t)buf[3] << 8) |
((size_t)buf[4] << 16) |
Optional: This looks rather ugly and uses C style casts. Doesn't this look like
it make use of function used to construct font table tags?

http://codereview.appspot.com/4659041/diff/1/src/pdf/SkPDFUtils.cpp
File src/pdf/SkPDFUtils.cpp (left):

http://codereview.appspot.com/4659041/diff/1/src/pdf/SkPDFUtils.cpp#oldcode137
src/pdf/SkPDFUtils.cpp:137: break;
Nit: Would it be better to leave the case statement and just remove the break
statement?

[Steve VanDeBogart, 2011-06-21 21:08:05]

http://codereview.appspot.com/4659041/diff/1/src/pdf/SkPDFFont.cpp
File src/pdf/SkPDFFont.cpp (right):

http://codereview.appspot.com/4659041/diff/1/src/pdf/SkPDFFont.cpp#newcode50
src/pdf/SkPDFFont.cpp:50: *size = (size_t)buf[2] | ((size_t)buf[3] << 8) |
((size_t)buf[4] << 16) |
On 2011/06/21 21:01:11, Chris Guillory wrote:
> Optional: This looks rather ugly and uses C style casts. Doesn't this look
like
> it make use of function used to construct font table tags?

It's similar to SkSetFourByteTag, but different enough that I wouldn't want to
reuse.  The casts are what fix the coverity complaint.

http://codereview.appspot.com/4659041/diff/1/src/pdf/SkPDFUtils.cpp
File src/pdf/SkPDFUtils.cpp (left):

http://codereview.appspot.com/4659041/diff/1/src/pdf/SkPDFUtils.cpp#oldcode137
src/pdf/SkPDFUtils.cpp:137: break;
On 2011/06/21 21:01:11, Chris Guillory wrote:
> Nit: Would it be better to leave the case statement and just remove the break
> statement?

The case statement would still be dead code because it can never happen.

[Chris Guillory, 2011-06-21 21:18:08]

SGTM.