Add m_pDocument in CPDF_Color and check if page date has been forced clear

Add m_pDocument in CPDF_Color and check if page date has been forced clear

This will prevent using freed pattern object.
This is a better solution than https://pdfium.googlesource.com/pdfium/+/1b9c5c4dc41956b8c5ab17b9a882adf8a2513768
and in essence revert that patch

BUG=409373
R=tsepez@chromium.org

Committed: https://pdfium.googlesource.com/pdfium/+/1d7dc1baba517bbf862e7d144e121b2ea4ffd33b

[Bo Xu, 2014-09-01 21:15:20]

bo_xu@foxitsoftware.com changed reviewers:
+ jun_fang@foxitsoftware.com, tsepez@chromium.org

[Bo Xu, 2014-09-01 21:15:20]

Hi Tom, please review this one, thanks!

[jun_fang, 2014-09-01 22:44:03]

https://codereview.chromium.org/522483003/diff/1/core/src/fpdfapi/fpdf_page/f...
File core/src/fpdfapi/fpdf_page/fpdf_page_colors.cpp (right):

https://codereview.chromium.org/522483003/diff/1/core/src/fpdfapi/fpdf_page/f...
core/src/fpdfapi/fpdf_page/fpdf_page_colors.cpp:1297: if
(!pPageData->IsForceClear()) {
need to check whether pPageData is NULL?

https://codereview.chromium.org/522483003/diff/1/core/src/fpdfapi/fpdf_page/f...
core/src/fpdfapi/fpdf_page/fpdf_page_colors.cpp:1357: if (pDocPageData &&
pDocPageData->IsForceClear()) {
should be "!pDocPageData->IsForceClear()"?

[Tom Sepez, 2014-09-02 17:15:50]

https://codereview.chromium.org/522483003/diff/20001/core/include/fpdfapi/fpd...
File core/include/fpdfapi/fpdf_resource.h (right):

https://codereview.chromium.org/522483003/diff/20001/core/include/fpdfapi/fpd...
core/include/fpdfapi/fpdf_resource.h:685: m_pBuffer = NULL;
Nit: can we clear these in the same order as which they are declared n the
class?

https://codereview.chromium.org/522483003/diff/20001/core/include/fpdfapi/fpd...
core/include/fpdfapi/fpdf_resource.h:725: 
nit: can this be protected/private?

https://codereview.chromium.org/522483003/diff/20001/core/include/fpdfapi/fpd...
core/include/fpdfapi/fpdf_resource.h:726: CPDF_Document*          m_pDocument;
Are we sure that the document will still be valid when this color is destructed?

[Tom Sepez, 2014-09-02 17:17:12]

https://codereview.chromium.org/522483003/diff/20001/core/include/fpdfapi/fpd...
File core/include/fpdfapi/fpdf_resource.h (right):

https://codereview.chromium.org/522483003/diff/20001/core/include/fpdfapi/fpd...
core/include/fpdfapi/fpdf_resource.h:726: CPDF_Document*          m_pDocument;
On 2014/09/02 17:15:49, Tom Sepez wrote:
> Are we sure that the document will still be valid when this color is
destructed?

In other words,
Are we sure that the document will still be valid up untl this color is
destructed?

[Bo Xu, 2014-09-02 19:05:34]

https://codereview.chromium.org/522483003/diff/20001/core/include/fpdfapi/fpd...
File core/include/fpdfapi/fpdf_resource.h (right):

https://codereview.chromium.org/522483003/diff/20001/core/include/fpdfapi/fpd...
core/include/fpdfapi/fpdf_resource.h:685: m_pBuffer = NULL;
On 2014/09/02 17:15:49, Tom Sepez wrote:
> Nit: can we clear these in the same order as which they are declared n the
> class?

Done.

https://codereview.chromium.org/522483003/diff/20001/core/include/fpdfapi/fpd...
core/include/fpdfapi/fpdf_resource.h:725: 
On 2014/09/02 17:15:49, Tom Sepez wrote:
> nit: can this be protected/private?

Done.

https://codereview.chromium.org/522483003/diff/20001/core/include/fpdfapi/fpd...
core/include/fpdfapi/fpdf_resource.h:726: CPDF_Document*          m_pDocument;
On 2014/09/02 17:15:49, Tom Sepez wrote:
> Are we sure that the document will still be valid when this color is
destructed?

Yes, the destrutor of color is in the destructor of CPDF_Document at line 96 of
fpdf_parser_document.cpp.

The lifetime of m_pDocument is until the very end at FPDF_CloseDocument()when
most of other resources has been released.

[Tom Sepez, 2014-09-02 19:09:33]

> The lifetime of m_pDocument is until the very end at FPDF_CloseDocument()when
> most of other resources has been released.
Excellent. LGTM.

[Bo Xu, 2014-09-02 19:13:34]

Committed patchset #3 (id:40001) manually as 1d7dc1b (presubmit successful).