Optimize QPDFParser for non-sparse QPDF_Arrays

Stop using nullptr to represent null objects.
Count null array elements and trigger creation of sparse arrays if null count is
greater than 100.

3 files changed, 29 insertions, 19 deletions

diff --git a/libqpdf/QPDFParser.cc b/libqpdf/QPDFParser.cc
index 09bf1601..4c43e487 100644
--- a/libqpdf/QPDFParser.cc
+++ b/libqpdf/QPDFParser.cc
@@ -27,16 +27,15 @@ namespace
     struct StackFrame
     {
         StackFrame(std::shared_ptr<InputSource> input) :
-            offset(input->tell()),
-            contents_string(""),
-            contents_offset(-1)
+            offset(input->tell())
         {
         }
 
         std::vector<std::shared_ptr<QPDFObject>> olist;
         qpdf_offset_t offset;
-        std::string contents_string;
-        qpdf_offset_t contents_offset;
+        std::string contents_string{""};
+        qpdf_offset_t contents_offset{-1};
+        int null_count{0};
     };
 } // namespace
 
@@ -50,6 +49,7 @@ QPDFParser::parse(bool& empty, bool content_stream)
     // this, it will cause a logic error to be thrown from
     // QPDF::inParse().
 
+    const static std::shared_ptr<QPDFObject> null_oh = QPDF_Null::create();
     QPDF::ParseGuard pg(context);
 
     empty = false;
@@ -67,7 +67,6 @@ QPDFParser::parse(bool& empty, bool content_stream)
     int good_count = 0;
     bool b_contents = false;
     bool is_null = false;
-    auto null_oh = QPDF_Null::create();
 
     while (!done) {
         bool bad = false;
@@ -156,6 +155,8 @@ QPDFParser::parse(bool& empty, bool content_stream)
 
         case QPDFTokenizer::tt_null:
             is_null = true;
+            ++frame.null_count;
+
             break;
 
         case QPDFTokenizer::tt_integer:
@@ -301,9 +302,11 @@ QPDFParser::parse(bool& empty, bool content_stream)
 
         case st_dictionary:
         case st_array:
-            if (!indirect_ref && !is_null) {
-                // No need to set description for direct nulls - they will
-                // become implicit.
+            if (is_null) {
+                object = null_oh;
+                // No need to set description for direct nulls - they probably
+                // will become implicit.
+            } else if (!indirect_ref) {
                 setDescription(object, input->getLastOffset());
             }
             set_offset = true;
@@ -326,7 +329,8 @@ QPDFParser::parse(bool& empty, bool content_stream)
             parser_state_e old_state = state_stack.back();
             state_stack.pop_back();
             if (old_state == st_array) {
-                object = QPDF_Array::create(std::move(olist));
+                object = QPDF_Array::create(
+                    std::move(olist), frame.null_count > 100);
                 setDescription(object, offset - 1);
                 // The `offset` points to the next of "[".  Set the rewind
                 // offset to point to the beginning of "[". This has been
@@ -381,7 +385,7 @@ QPDFParser::parse(bool& empty, bool content_stream)
                     // Calculate value.
                     std::shared_ptr<QPDFObject> val;
                     if (iter != olist.end()) {
-                        val = *iter ? *iter : QPDF_Null::create();
+                        val = *iter;
                         ++iter;
                     } else {
                         QTC::TC("qpdf", "QPDFParser no val for last key");
diff --git a/libqpdf/QPDF_Array.cc b/libqpdf/QPDF_Array.cc
index 7a3a29ca..31abd8eb 100644
--- a/libqpdf/QPDF_Array.cc
+++ b/libqpdf/QPDF_Array.cc
@@ -11,15 +11,19 @@ QPDF_Array::QPDF_Array(std::vector<QPDFObjectHandle> const& v) :
     setFromVector(v);
 }
 
-QPDF_Array::QPDF_Array(std::vector<std::shared_ptr<QPDFObject>>&& v) :
-    QPDFValue(::ot_array, "array")
+QPDF_Array::QPDF_Array(
+    std::vector<std::shared_ptr<QPDFObject>>&& v, bool sparse) :
+    QPDFValue(::ot_array, "array"),
+    sparse(sparse)
 {
     setFromVector(std::move(v));
 }
 
 QPDF_Array::QPDF_Array(SparseOHArray const& items) :
     QPDFValue(::ot_array, "array"),
+    sparse(true),
     sp_elements(items)
+
 {
 }
 
@@ -37,9 +41,10 @@ QPDF_Array::create(std::vector<QPDFObjectHandle> const& items)
 }
 
 std::shared_ptr<QPDFObject>
-QPDF_Array::create(std::vector<std::shared_ptr<QPDFObject>>&& items)
+QPDF_Array::create(
+    std::vector<std::shared_ptr<QPDFObject>>&& items, bool sparse)
 {
-    return do_create(new QPDF_Array(std::move(items)));
+    return do_create(new QPDF_Array(std::move(items), sparse));
 }
 
 std::shared_ptr<QPDFObject>
@@ -196,8 +201,9 @@ QPDF_Array::setFromVector(std::vector<std::shared_ptr<QPDFObject>>&& v)
     if (sparse) {
         sp_elements = SparseOHArray();
         for (auto&& item: v) {
-            if (item) {
-                sp_elements.append(item);
+            if (item->getTypeCode() != ::ot_null ||
+                item->getObjGen().isIndirect()) {
+                sp_elements.append(std::move(item));
             } else {
                 ++sp_elements.n_elements;
             }
diff --git a/libqpdf/qpdf/QPDF_Array.hh b/libqpdf/qpdf/QPDF_Array.hh
index 1c4227ba..00c7f59d 100644
--- a/libqpdf/qpdf/QPDF_Array.hh
+++ b/libqpdf/qpdf/QPDF_Array.hh
@@ -15,7 +15,7 @@ class QPDF_Array: public QPDFValue
     static std::shared_ptr<QPDFObject>
     create(std::vector<QPDFObjectHandle> const& items);
     static std::shared_ptr<QPDFObject>
-    create(std::vector<std::shared_ptr<QPDFObject>>&& items);
+    create(std::vector<std::shared_ptr<QPDFObject>>&& items, bool sparse);
     static std::shared_ptr<QPDFObject> create(SparseOHArray const& items);
     static std::shared_ptr<QPDFObject> create(OHArray const& items);
     virtual std::shared_ptr<QPDFObject> copy(bool shallow = false);
@@ -36,7 +36,7 @@ class QPDF_Array: public QPDFValue
 
   private:
     QPDF_Array(std::vector<QPDFObjectHandle> const& items);
-    QPDF_Array(std::vector<std::shared_ptr<QPDFObject>>&& items);
+    QPDF_Array(std::vector<std::shared_ptr<QPDFObject>>&& items, bool sparse);
     QPDF_Array(SparseOHArray const& items);
     QPDF_Array(OHArray const& items);
     bool sparse{false};