Move QPDF::ObjCache::end_before_space etc to Xref_table

Also, delay adjustments for compressed objects until needed by linearization checks.

Move QPDF::ObjCache::end_before_space etc to Xref_table
Also, delay adjustments for compressed objects until needed by linearization checks.
m-holger
1 parent 28c13f54
Showing 4 changed files with 93 additions and 62 deletions
include/qpdf/QPDF.hh
libqpdf/QPDF.cc
libqpdf/QPDF_linearization.cc
libqpdf/qpdf/QPDF_private.hh
@@ -786,11 +786,7 @@ class QPDF
     std::shared_ptr<QPDFObject> getObjectForParser(int id, int gen, bool parse_pdf);
     std::shared_ptr<QPDFObject> getObjectForJSON(int id, int gen);
     void removeObject(QPDFObjGen og);
-    void updateCache(
-        QPDFObjGen const& og,
-        std::shared_ptr<QPDFObject> const& object,
-        qpdf_offset_t end_before_space,
-        qpdf_offset_t end_after_space);
+    void updateCache(QPDFObjGen const& og, std::shared_ptr<QPDFObject> const& object);
     static QPDFExc damagedPDF(
         InputSource& input,
         std::string const& object,
@@ -1953,7 +1953,9 @@ QPDF::readObjectAtOffset(
             // could use !check_og in place of skip_cache_if_in_xref.
             QTC::TC("qpdf", "QPDF skipping cache for known unchecked object");
         } else {
-            updateCache(og, oh.getObj(), end_before_space, end_after_space);
+            m->xref_table.linearization_offsets(
+                toS(og.getObj()), end_before_space, end_after_space);
+            updateCache(og, oh.getObj());
         }
     }
@@ -1972,7 +1974,7 @@ QPDF::resolve(QPDFObjGen og)
         // has to be resolved during object parsing, such as stream length.
         QTC::TC("qpdf", "QPDF recursion loop in resolve");
         warn(damagedPDF("", "loop detected resolving object " + og.unparse(' ')));
-        updateCache(og, QPDF_Null::create(), -1, -1);
+        updateCache(og, QPDF_Null::create());
         return m->obj_cache[og].object.get();
     }
     ResolveRecorder rr(this, og);
@@ -2008,7 +2010,7 @@ QPDF::resolve(QPDFObjGen og)
     if (isUnresolved(og)) {
         // PDF spec says unknown objects resolve to the null object.
         QTC::TC("qpdf", "QPDF resolve failure to null");
-        updateCache(og, QPDF_Null::create(), -1, -1);
+        updateCache(og, QPDF_Null::create());
     }
     auto result(m->obj_cache[og].object);
@@ -2030,12 +2032,6 @@ QPDF::resolveObjectsInStream(int obj_stream_number)
             "supposed object stream " + std::to_string(obj_stream_number) + " is not a stream");
     }
-    // For linearization data in the object, use the data from the object stream for the objects in
-    // the stream.
-    QPDFObjGen stream_og(obj_stream_number, 0);
-    qpdf_offset_t end_before_space = m->obj_cache[stream_og].end_before_space;
-    qpdf_offset_t end_after_space = m->obj_cache[stream_og].end_after_space;
-
     QPDFObjectHandle dict = obj_stream.getDict();
     if (!dict.isDictionaryOfType("/ObjStm")) {
         QTC::TC("qpdf", "QPDF ERR object stream with wrong type");
@@ -2101,7 +2097,7 @@ QPDF::resolveObjectsInStream(int obj_stream_number)
             int offset = iter.second;
             input->seek(offset, SEEK_SET);
             QPDFObjectHandle oh = readObjectInStream(input, iter.first);
-            updateCache(og, oh.getObj(), end_before_space, end_after_space);
+            updateCache(og, oh.getObj());
         } else {
             QTC::TC("qpdf", "QPDF not caching overridden objstm object");
         }
@@ -2116,20 +2112,14 @@ QPDF::newIndirect(QPDFObjGen const&amp; og, std::shared_ptr&lt;QPDFObject&gt; const&amp; obj)
 }
 void
-QPDF::updateCache(
-    QPDFObjGen const& og,
-    std::shared_ptr<QPDFObject> const& object,
-    qpdf_offset_t end_before_space,
-    qpdf_offset_t end_after_space)
+QPDF::updateCache(QPDFObjGen const& og, std::shared_ptr<QPDFObject> const& object)
 {
     object->setObjGen(this, og);
     if (isCached(og)) {
         auto& cache = m->obj_cache[og];
         cache.object->assign(object);
-        cache.end_before_space = end_before_space;
-        cache.end_after_space = end_after_space;
     } else {
-        m->obj_cache[og] = ObjCache(object, end_before_space, end_after_space);
+        m->obj_cache[og] = ObjCache(object);
     }
 }
@@ -2159,7 +2149,7 @@ QPDFObjectHandle
 QPDF::makeIndirectFromQPDFObject(std::shared_ptr<QPDFObject> const& obj)
 {
     QPDFObjGen next{nextObjGen()};
-    m->obj_cache[next] = ObjCache(obj, -1, -1);
+    m->obj_cache[next] = ObjCache(obj);
     return newIndirect(next, m->obj_cache[next].object);
 }
@@ -2246,7 +2236,7 @@ QPDF::getObject(QPDFObjGen const&amp; og)
     } else if (m->xref_table.initialized() && !m->xref_table.type(og)) {
         return QPDF_Null::create();
     } else {
-        auto result = m->obj_cache.try_emplace(og, QPDF_Unresolved::create(this, og), -1, -1);
+        auto result = m->obj_cache.try_emplace(og, QPDF_Unresolved::create(this, og));
         return {result.first->second.object};
     }
 }
@@ -2282,7 +2272,7 @@ QPDF::replaceObject(QPDFObjGen const&amp; og, QPDFObjectHandle oh)
         QTC::TC("qpdf", "QPDF replaceObject called with indirect object");
         throw std::logic_error("QPDF::replaceObject called with indirect object handle");
     }
-    updateCache(og, oh.getObj(), -1, -1);
+    updateCache(og, oh.getObj());
 }
 void
@@ -288,9 +288,8 @@ QPDF::readHintStream(Pipeline&amp; pl, qpdf_offset_t offset, size_t length)
     QPDFObjGen og;
     QPDFObjectHandle H =
         readObjectAtOffset(false, offset, "linearization hint stream", QPDFObjGen(0, 0), og, false);
-    ObjCache& oc = m->obj_cache[og];
-    qpdf_offset_t min_end_offset = oc.end_before_space;
-    qpdf_offset_t max_end_offset = oc.end_after_space;
+    qpdf_offset_t min_end_offset = m->xref_table.end_before_space(og);
+    qpdf_offset_t max_end_offset = m->xref_table.end_after_space(og);
     if (!H.isStream()) {
         throw damagedPDF("linearization dictionary", "hint table is not a stream");
     }
@@ -301,14 +300,11 @@ QPDF::readHintStream(Pipeline&amp; pl, qpdf_offset_t offset, size_t length)
     // increasing length to cover it, even though the specification says all objects in the
     // linearization parameter dictionary must be direct.  We have to get the file position of the
     // end of length in this case.
-    QPDFObjectHandle length_obj = Hdict.getKey("/Length");
-    if (length_obj.isIndirect()) {
+    auto length_og = Hdict.getKey("/Length").getObjGen();
+    if (length_og.isIndirect()) {
         QTC::TC("qpdf", "QPDF hint table length indirect");
-        // Force resolution
-        (void)length_obj.getIntValue();
-        ObjCache& oc2 = m->obj_cache[length_obj.getObjGen()];
-        min_end_offset = oc2.end_before_space;
-        max_end_offset = oc2.end_after_space;
+        min_end_offset = m->xref_table.end_before_space(length_og);
+        max_end_offset = m->xref_table.end_after_space(length_og);
     } else {
         QTC::TC("qpdf", "QPDF hint table length direct");
     }
@@ -503,13 +499,14 @@ QPDF::checkLinearizationInternal()
     qpdf_offset_t max_E = -1;
     for (auto const& oh: m->part6) {
         QPDFObjGen og(oh.getObjGen());
-        if (m->obj_cache.count(og) == 0) {
+        auto before = m->xref_table.end_before_space(og);
+        auto after = m->xref_table.end_after_space(og);
+        if (before <= 0) {
             // All objects have to have been dereferenced to be classified.
             throw std::logic_error("linearization part6 object not in cache");
         }
-        ObjCache const& oc = m->obj_cache[og];
-        min_E = std::max(min_E, oc.end_before_space);
-        max_E = std::max(max_E, oc.end_after_space);
+        min_E = std::max(min_E, before);
+        max_E = std::max(max_E, after);
     }
     if ((p.first_page_end < min_E) || (p.first_page_end > max_E)) {
         QTC::TC("qpdf", "QPDF warn /E mismatch");
@@ -536,10 +533,11 @@ QPDF::maxEnd(ObjUser const&amp; ou)
     }
     qpdf_offset_t end = 0;
     for (auto const& og: m->obj_user_to_objects[ou]) {
-        if (m->obj_cache.count(og) == 0) {
+        auto e = m->xref_table.end_after_space(og);
+        if (e <= 0) {
             stopOnError("unknown object referenced in object user table");
         }
-        end = std::max(end, m->obj_cache[og].end_after_space);
+        end = std::max(end, e);
     }
     return end;
 }
@@ -599,15 +597,13 @@ QPDF::lengthNextN(int first_object, int n)
     int length = 0;
     for (int i = 0; i < n; ++i) {
         QPDFObjGen og(first_object + i, 0);
-        if (m->xref_table.type(og) == 0) {
+        auto end = m->xref_table.end_after_space(og);
+        if (end <= 0) {
             linearizationWarning(
                 "no xref table entry for " + std::to_string(first_object + i) + " 0");
-        } else {
-            if (m->obj_cache.count(og) == 0) {
-                stopOnError("found unknown object while calculating length for linearization data");
-            }
-            length += toI(m->obj_cache[og].end_after_space - getLinearizationOffset(og));
+            continue;
         }
+        length += toI(end - getLinearizationOffset(og));
     }
     return length;
 }
@@ -151,6 +151,49 @@ class QPDF::Xref_table
     // For Linearization
+    qpdf_offset_t
+    end_after_space(QPDFObjGen og)
+    {
+        auto& e = entry(toS(og.getObj()));
+        switch (e.type()) {
+        case 1:
+            return e.end_after_space_;
+        case 2:
+            {
+                auto es = entry(toS(e.stream_number()));
+                return es.type() == 1 ? es.end_after_space_ : 0;
+            }
+        default:
+            return 0;
+        }
+    }
+
+    qpdf_offset_t
+    end_before_space(QPDFObjGen og)
+    {
+        auto& e = entry(toS(og.getObj()));
+        switch (e.type()) {
+        case 1:
+            return e.end_before_space_;
+        case 2:
+            {
+                auto es = entry(toS(e.stream_number()));
+                return es.type() == 1 ? es.end_before_space_ : 0;
+            }
+        default:
+            return 0;
+        }
+    }
+
+    void
+    linearization_offsets(size_t id, qpdf_offset_t before, qpdf_offset_t after)
+    {
+        if (type(id)) {
+            table[id].end_before_space_ = before;
+            table[id].end_after_space_ = after;
+        }
+    }
+
     bool
     uncompressed_after_compressed() const noexcept
     {
@@ -192,6 +235,14 @@ class QPDF::Xref_table
     struct Entry
     {
+        Entry() = default;
+
+        Entry(int gen, Xref entry) :
+            gen_(gen),
+            entry(entry)
+        {
+        }
+
         int
         gen() const noexcept
         {
@@ -224,8 +275,16 @@ class QPDF::Xref_table
         int gen_{0};
         Xref entry;
+        qpdf_offset_t end_before_space_{0};
+        qpdf_offset_t end_after_space_{0};
     };
+    Entry&
+    entry(size_t id)
+    {
+        return id < table.size() ? table[id] : table[0];
+    }
+
     void read(qpdf_offset_t offset);
     // Methods to parse tables
@@ -384,24 +443,14 @@ class QPDF::Pipe
 class QPDF::ObjCache
 {
   public:
-    ObjCache() :
-        end_before_space(0),
-        end_after_space(0)
-    {
-    }
-    ObjCache(
-        std::shared_ptr<QPDFObject> object,
-        qpdf_offset_t end_before_space = 0,
-        qpdf_offset_t end_after_space = 0) :
-        object(object),
-        end_before_space(end_before_space),
-        end_after_space(end_after_space)
+    ObjCache() = default;
+
+    ObjCache(std::shared_ptr<QPDFObject> object) :
+        object(object)
     {
     }
     std::shared_ptr<QPDFObject> object;
-    qpdf_offset_t end_before_space;
-    qpdf_offset_t end_after_space;
 };
 class QPDF::ObjCopier