Fix blank page issue caused by too strict correction on bpc
authorJun Fang <jun_fang@foxitsoftware.com>
Tue, 29 Sep 2015 02:24:54 +0000 (10:24 +0800)
committerJun Fang <jun_fang@foxitsoftware.com>
Tue, 29 Sep 2015 02:24:54 +0000 (10:24 +0800)
For bit per component (bpc), PDF spec mentions that a RunLengthDecode or DCTDecode filter shall always deliver 8-bit samples. However, some PDF files don't follow this rule. We can find that filter is RunLengthDecode but bpc is 1 in the provided test file. In this case, pdfium will correct bpc to 8 but the actual bpc is 1. It causes a failure because the data is much more than the expected. To handle this case, pdfium doesn't correct bpc to 8 when the original bpc is 1.

BUG=512557
R=tsepez@chromium.org

Review URL: https://codereview.chromium.org/1328213002 .

core/src/fpdfapi/fpdf_render/fpdf_render_loadimage.cpp
testing/resources/pixel/bug_512557.in [new file with mode: 0644]
testing/resources/pixel/bug_512557_expected.pdf.0.png [new file with mode: 0644]

index 0362ff2..9497943 100644 (file)
@@ -957,8 +957,11 @@ void CPDF_DIBSource::ValidateDictParam() {
         m_bpc = 1;
         m_nComponents = 1;
       }
-      if (filter == FX_BSTRC("RunLengthDecode") ||
-          filter == FX_BSTRC("DCTDecode")) {
+      if (filter == FX_BSTRC("RunLengthDecode")) {
+        if (m_bpc != 1) {
+          m_bpc = 8;
+        }
+      } else if (filter == FX_BSTRC("DCTDecode")) {
         m_bpc = 8;
       }
     } else if (pFilter->GetType() == PDFOBJ_ARRAY) {
diff --git a/testing/resources/pixel/bug_512557.in b/testing/resources/pixel/bug_512557.in
new file mode 100644 (file)
index 0000000..5f35334
Binary files /dev/null and b/testing/resources/pixel/bug_512557.in differ
diff --git a/testing/resources/pixel/bug_512557_expected.pdf.0.png b/testing/resources/pixel/bug_512557_expected.pdf.0.png
new file mode 100644 (file)
index 0000000..66c73ae
Binary files /dev/null and b/testing/resources/pixel/bug_512557_expected.pdf.0.png differ