Fix some bugs in StructTreeRoot parsing of parent tree

- Add support for parsing child nodes in the number tree
- Number tree keys do not have to be consecutive numbers. Use
  map instead of vector for parentTree.
- Due to performance impact of iterating a map instead of
  vector in parentTreeAdd, add a reverse mapping from Ref
  to parentTree.
- Add mcid parameter to findParentElement() to enable finding
  the parent when there are multiple MCIDs on the same page.
- Move RefCompare from pdfinfo.cc to Object.h so it can be
  used by other files.

Bug #103912

1 files changed, 10 insertions, 4 deletions

diff --git a/poppler/StructTreeRoot.h b/poppler/StructTreeRoot.h
index 3b1f3c84..ca688499 100644
--- a/poppler/StructTreeRoot.h
+++ b/poppler/StructTreeRoot.h
@@ -18,6 +18,7 @@
 #include "goo/gtypes.h"
 #include "Object.h"
 #include "StructElement.h"
+#include <map>
 #include <vector>
 
 class Dict;
@@ -43,9 +44,12 @@ public:
     }
   }
 
-  const StructElement *findParentElement(unsigned index) const {
-    if (index < parentTree.size() && parentTree[index].size() == 1) {
-      return parentTree[index][0].element;
+  const StructElement *findParentElement(int key, unsigned mcid = 0) const {
+    auto it = parentTree.find(key);
+    if (it != parentTree.end()) {
+      if (mcid < it->second.size()) {
+	return it->second[mcid].element;
+      }
     }
     return NULL;
   }
@@ -71,9 +75,11 @@ private:
   Object roleMap;
   Object classMap;
   ElemPtrArray elements;
-  std::vector< std::vector<Parent> > parentTree;
+  std::map<int, std::vector<Parent> > parentTree;
+  std::multimap<Ref, Parent*, RefCompare> refToParentMap;
 
   void parse(Dict *rootDict);
+  void parseNumberTreeNode(Dict *node);
   void parentTreeAdd(const Ref &objectRef, StructElement *element);
 
   friend class StructElement;