Embedpdf/feature/attachment capture

core/fpdfapi/parser/cpdf_base_document.h

@@ -47,6 +47,11 @@ class CPDF_BaseDocument final : public CPDF_Document, public Retainable {
   // full pass over the file — unless the embedder supplied it through
   // SetKnownRawBaseSha256(). Reads all-zero when the file cannot be read.
   const std::array<uint8_t, 32>& GetRawBaseSha256() const;
+  // The same hash when it is known already (supplied, or worked out once),
+  // else null: never hashes the file.
+  const std::array<uint8_t, 32>* GetKnownRawBaseSha256() const {
+    return raw_base_sha256_ ? &*raw_base_sha256_ : nullptr;
+  }
   // A host that already hashed the exact bytes it handed us (a storage layer
   // verifying a download) need not pay for a second pass. Only an identity
   // claim: a wrong value makes this host's own artifacts fail to open on the

core/fpdfdoc/cpdf_bookmarktree.cpp

@@ -17,11 +17,44 @@ CPDF_BookmarkTree::CPDF_BookmarkTree(const CPDF_Document* doc)
 
 CPDF_BookmarkTree::~CPDF_BookmarkTree() = default;
 
+// EmbedPDF(bug-fix): see the declaration. An item has no /Type, and the
+// outline's dictionary may say /Outlines; the other keys belong to pages and
+// the page tree (/Kids /Contents /MediaBox /Resources), annotations and
+// fields (/Subtype /FT) and actions (/S, whose /Next chains actions).
+// static
+bool CPDF_BookmarkTree::IsOutlineNode(const CPDF_Dictionary* dict) {
+  if (!dict) {
+    return false;
+  }
+  if (dict->KeyExist("Type") && dict->GetNameFor("Type") != "Outlines") {
+    return false;
+  }
+  for (const char* key :
+       {"Subtype", "FT", "S", "Kids", "Contents", "MediaBox", "Resources"}) {
+    if (dict->KeyExist(key)) {
+      return false;
+    }
+  }
+  return true;
+}
+
+namespace {
+
+// EmbedPDF(bug-fix): |dict| as a bookmark when IsOutlineNode() takes it.
+CPDF_Bookmark OutlineItem(RetainPtr<const CPDF_Dictionary> dict) {
+  return CPDF_BookmarkTree::IsOutlineNode(dict.Get())
+             ? CPDF_Bookmark(std::move(dict))
+             : CPDF_Bookmark();
+}
+
+}  // namespace
+
 CPDF_Bookmark CPDF_BookmarkTree::GetFirstChild(
     const CPDF_Bookmark& parent) const {
   const CPDF_Dictionary* parent_dict = parent.GetDict();
   if (parent_dict) {
-    return CPDF_Bookmark(parent_dict->GetDictFor("First"));
+    // EmbedPDF(bug-fix): only an item IsOutlineNode() takes.
+    return OutlineItem(parent_dict->GetDictFor("First"));
   }
 
   const CPDF_Dictionary* root = document_->GetRoot();
@@ -31,8 +64,10 @@ CPDF_Bookmark CPDF_BookmarkTree::GetFirstChild(
 
   RetainPtr<const CPDF_Dictionary> outlines =
       root->GetDictFor(pdfium::catalog::kOutlines);
-  return outlines ? CPDF_Bookmark(outlines->GetDictFor("First"))
-                  : CPDF_Bookmark();
+  // EmbedPDF(bug-fix): only an outline and item IsOutlineNode() takes.
+  return IsOutlineNode(outlines.Get())
+             ? OutlineItem(outlines->GetDictFor("First"))
+             : CPDF_Bookmark();
 }
 
 CPDF_Bookmark CPDF_BookmarkTree::GetNextSibling(
@@ -43,5 +78,6 @@ CPDF_Bookmark CPDF_BookmarkTree::GetNextSibling(
   }
 
   RetainPtr<const CPDF_Dictionary> next = dict->GetDictFor("Next");
-  return next != dict ? CPDF_Bookmark(std::move(next)) : CPDF_Bookmark();
+  // EmbedPDF(bug-fix): only an item IsOutlineNode() takes.
+  return next != dict ? OutlineItem(std::move(next)) : CPDF_Bookmark();
 }

core/fpdfdoc/cpdf_bookmarktree.h

@@ -10,13 +10,22 @@
 #include "core/fpdfdoc/cpdf_bookmark.h"
 #include "core/fxcrt/unowned_ptr.h"
 
+// EmbedPDF(bug-fix): IsOutlineNode() below reads one.
+class CPDF_Dictionary;
 class CPDF_Document;
 
 class CPDF_BookmarkTree {
  public:
   explicit CPDF_BookmarkTree(const CPDF_Document* doc);
   ~CPDF_BookmarkTree();
 
+  // EmbedPDF(bug-fix): upstream took any dictionary that /Outlines, /First or
+  // /Next reaches as part of the outline, so a broken file that links a page,
+  // a field or an action there showed it as a bookmark, and an outline
+  // writer rewrote its links. Whether |dict| can be the outline's dictionary
+  // or one of its items: it has no key that gives a dictionary another role.
+  static bool IsOutlineNode(const CPDF_Dictionary* dict);
+
   CPDF_Bookmark GetFirstChild(const CPDF_Bookmark& parent) const;
   CPDF_Bookmark GetNextSibling(const CPDF_Bookmark& bookmark) const;
 

fpdfsdk/epdf_c_api_test.c

@@ -217,6 +217,10 @@ int CheckEmbedPDFCApi(void) {
   CHK(EPDFDoc_GetAttachmentKey);
 
   // epdf_capture.h
+  CHK(EPDFAttachment_ExportCanonicalToOwnedBuffer);
+  CHK(EPDFAttachment_ExportRawToOwnedBuffer);
+  CHK(EPDFAttachment_ImportRaw);
+  CHK(EPDFDoc_ExportDictCanonicalToOwnedBuffer);
   CHK(EPDFDoc_ExportDictRawToOwnedBuffer);
   CHK(EPDFDoc_ImportDictRaw);
   CHK(EPDFForm_ExportFieldRawToOwnedBuffer);