Merge pull request #1277 from m-holger/objtable

Refactor ObjTable

Merge pull request #1277 from m-holger/objtable
Refactor ObjTable
m-holger · GitHub
2 parents 7f3b05dd 7777ea84
Showing 4 changed files with 112 additions and 20 deletions
libqpdf/QPDFWriter.cc
libqpdf/qpdf/ObjTable.hh
libtests/obj_table.cc
libtests/qtest/obj_table/obj_table.out
@@ -2077,8 +2077,8 @@ void
 QPDFWriter::initializeTables(size_t extra)
 {
     auto size = QIntC::to_size(QPDF::Writer::tableSize(m->pdf) + 100) + extra;
-    m->obj.initialize(size);
-    m->new_obj.initialize(size);
+    m->obj.resize(size);
+    m->new_obj.resize(size);
 }
 void
@@ -28,6 +28,7 @@ template &lt;class T&gt;
 class ObjTable: public std::vector<T>
 {
   public:
+    using reference = T&;
     ObjTable() = default;
     ObjTable(const ObjTable&) = delete;
     ObjTable(ObjTable&&) = delete;
@@ -99,17 +100,30 @@ class ObjTable: public std::vector&lt;T&gt;
         return element(id);
     }
+    // emplace_back to the end of the vector. If there are any conflicting sparse elements, emplace
+    // them to the back of the vector before adding the current element.
+    template <class... Args>
+    inline T&
+    emplace_back(Args&&... args)
+    {
+        if (min_sparse == std::vector<T>::size()) {
+            return emplace_back_large(std::forward<Args&&...>(args...));
+        }
+        return std::vector<T>::emplace_back(std::forward<Args&&...>(args...));
+    }
+
     void
-    initialize(size_t idx)
-    {
-        if (std::vector<T>::size() > 0 || sparse_elements.size() > 0) {
-            throw ::std::logic_error("ObjTable accessed before initialization");
-        } else if (
-            idx >= static_cast<size_t>(std::numeric_limits<int>::max()) ||
-            idx >= std::vector<T>::max_size()) {
-            throw std::runtime_error("Invalid maximum object id initializing ObjTable.");
-        } else {
-            std::vector<T>::resize(++idx);
+    resize(size_t a_size)
+    {
+        std::vector<T>::resize(a_size);
+        if (a_size > min_sparse) {
+            auto it = sparse_elements.begin();
+            auto end = sparse_elements.end();
+            while (it != end && it->first < a_size) {
+                std::vector<T>::operator[](it->first) = std::move(it->second);
+                it = sparse_elements.erase(it);
+            }
+            min_sparse = (it == end) ? std::numeric_limits<size_t>::max() : it->first;
         }
     }
@@ -127,30 +141,62 @@ class ObjTable: public std::vector&lt;T&gt;
   private:
     std::map<size_t, T> sparse_elements;
+    // Smallest id present in sparse_elements.
+    size_t min_sparse{std::numeric_limits<size_t>::max()};
     inline T&
     element(size_t idx)
     {
         if (idx < std::vector<T>::size()) {
             return std::vector<T>::operator[](idx);
-        } else if (idx < static_cast<size_t>(std::numeric_limits<int>::max())) {
+        }
+        return large_element(idx);
+    }
+
+    // Must only be called by element. Separated out from element to keep inlined code tight.
+    T&
+    large_element(size_t idx)
+    {
+        static const size_t max_size = std::vector<T>::max_size();
+        if (idx < min_sparse) {
+            min_sparse = idx;
+        }
+        if (idx < max_size) {
             return sparse_elements[idx];
         }
-        throw std::runtime_error("Invalid object id accessing ObjTable.");
+        throw std::runtime_error("Impossibly large object id encountered accessing ObjTable");
         return element(0); // doesn't return
     }
     inline T const&
     element(size_t idx) const
     {
+        static const size_t max_size = std::vector<T>::max_size();
         if (idx < std::vector<T>::size()) {
             return std::vector<T>::operator[](idx);
-        } else if (idx < static_cast<size_t>(std::numeric_limits<int>::max())) {
+        }
+        if (idx < max_size) {
             return sparse_elements.at(idx);
         }
-        throw std::runtime_error("Invalid object id accessing ObjTable.");
+        throw std::runtime_error("Impossibly large object id encountered accessing ObjTable");
         return element(0); // doesn't return
     }
+
+    // Must only be called by emplace_back. Separated out from emplace_back to keep inlined code
+    // tight.
+    template <class... Args>
+    T&
+    emplace_back_large(Args&&... args)
+    {
+        auto it = sparse_elements.begin();
+        auto end = sparse_elements.end();
+        while (it != end && it->first == std::vector<T>::size()) {
+            std::vector<T>::emplace_back(std::move(it->second));
+            it = sparse_elements.erase(it);
+        }
+        min_sparse = (it == end) ? std::numeric_limits<size_t>::max() : it->first;
+        return std::vector<T>::emplace_back(std::forward<Args&&...>(args...));
+    }
 };
 #endif // OBJTABLE_HH
@@ -2,6 +2,11 @@
 struct Test
 {
+    Test() = default;
+    Test(int value) :
+        value(value)
+    {
+    }
     int value{0};
 };
@@ -10,7 +15,7 @@ class Table: public ObjTable&lt;Test&gt;
   public:
     Table()
     {
-        initialize(5);
+        resize(5);
     }
     void
@@ -20,9 +25,23 @@ class Table: public ObjTable&lt;Test&gt;
             (*this)[i].value = 2 * i;
             (*this)[1000 + i].value = 2 * (1000 + i);
         }
+        for (int i: {50, 60, 70, 98, 99, 100, 101, 150, 198, 199, 200, 201}) {
+            (*this)[i].value = 2 * i;
+        }
+        resize(100);
+        for (int i: {1, 99, 100, 105, 110, 120, 205, 206, 207, 210}) {
+            (*this)[i].value = 3 * i;
+        }
+        resize(200);
+
+        for (int i = 1; i < 10; ++i) {
+            emplace_back(i);
+        }
         forEach([](auto i, auto const& item) -> void {
-            std::cout << std::to_string(i) << " : " << std::to_string(item.value) << "\n";
+            if (item.value) {
+                std::cout << std::to_string(i) << " : " << std::to_string(item.value) << "\n";
+            }
         });
         std::cout << "2000 : " << std::to_string((*this)[2000].value) << "\n";
-0 : 0
-1 : 2
+1 : 3
 2 : 4
 3 : 6
 4 : 8
@@ -8,6 +7,34 @@
 7 : 14
 8 : 16
 9 : 18
+50 : 100
+60 : 120
+70 : 140
+98 : 196
+99 : 297
+100 : 300
+101 : 202
+105 : 315
+110 : 330
+120 : 360
+150 : 300
+198 : 396
+199 : 398
+200 : 400
+201 : 402
+202 : 1
+203 : 2
+204 : 3
+205 : 615
+206 : 618
+207 : 621
+208 : 4
+209 : 5
+210 : 630
+211 : 6
+212 : 7
+213 : 8
+214 : 9
 1000 : 2000
 1001 : 2002
 1002 : 2004