summaryrefslogtreecommitdiff
path: root/core/include/fpdfapi
diff options
context:
space:
mode:
authorDan Sinclair <dsinclair@chromium.org>2015-10-26 16:25:38 -0400
committerDan Sinclair <dsinclair@chromium.org>2015-10-26 16:25:38 -0400
commit95d25e4585358c74ae91c2ed5e08099ebbfdf24c (patch)
tree89207598714eda2da04af966a013c905a1c63dcf /core/include/fpdfapi
parent85361b227ad6786d2aeef8409b79a8d077a26ee9 (diff)
downloadpdfium-95d25e4585358c74ae91c2ed5e08099ebbfdf24c.tar.xz
Revert "Make m_pShadingObj a CPDF_Stream instead of CPDF_Object."
There appear to be a bunch of corpus tests which fail with this change such as FAILURE: shading1.pdf This reverts commit 85361b227ad6786d2aeef8409b79a8d077a26ee9. Make m_pShadingObj a CPDF_Stream instead of CPDF_Object. This object is required to be a stream and was being converted as such. With the new type checking this caused us to pass a nullptr where previously we'd have, incorrectly, cast a CPDF_Dictionary to a CPDF_Stream. This CL changes the m_pShadingObj to always be a CPDF_Stream. Then, we never go down the bad code path because we check if m_pShadingObj is nullptr earlier and bail out. BUG=chromium:547706 TBR=tsepez@chromium.org Review URL: https://codereview.chromium.org/1414283003 .
Diffstat (limited to 'core/include/fpdfapi')
-rw-r--r--core/include/fpdfapi/fpdf_resource.h2
1 files changed, 1 insertions, 1 deletions
diff --git a/core/include/fpdfapi/fpdf_resource.h b/core/include/fpdfapi/fpdf_resource.h
index 0b85333cfc..9fe6e2b95f 100644
--- a/core/include/fpdfapi/fpdf_resource.h
+++ b/core/include/fpdfapi/fpdf_resource.h
@@ -688,7 +688,7 @@ class CPDF_ShadingPattern : public CPDF_Pattern {
~CPDF_ShadingPattern() override;
- CPDF_Stream* m_pShadingObj;
+ CPDF_Object* m_pShadingObj;
FX_BOOL m_bShadingObj;