Add new private method QPDF::readStream

include/qpdf/QPDF.hh

@@ -1008,6 +1008,7 @@ class QPDF
     void setLastObjectDescription(std::string const& description, QPDFObjGen const& og);
     QPDFObjectHandle readTrailer();
     QPDFObjectHandle readObject(std::string const& description, QPDFObjGen og);
+    void readStream(QPDFObjectHandle& object, QPDFObjGen og, qpdf_offset_t offset);
     QPDFObjectHandle readObjectInStream(std::shared_ptr<InputSource>, QPDFObjGen og);
     size_t recoverStreamLength(
         std::shared_ptr<InputSource> input, QPDFObjGen const& og, qpdf_offset_t stream_offset);

libqpdf/QPDF.cc

@@ -1296,7 +1296,7 @@ QPDF::readObject(std::string const& description, QPDFObjGen og)
     std::shared_ptr<StringDecrypter> decrypter_ph;
     StringDecrypter* decrypter = nullptr;
     if (m->encp->encrypted) {
-        decrypter_ph = std::make_shared<StringDecrypter>(this, og);
+        decrypter_ph = std::make_unique<StringDecrypter>(this, og);
         decrypter = decrypter_ph.get();
     }
     auto object = QPDFParser(m->file, m->last_object_description, m->tokenizer, decrypter, this)
@@ -1309,20 +1309,35 @@ QPDF::readObject(std::string const& description, QPDFObjGen og)
         // check for stream
         qpdf_offset_t cur_offset = m->file->tell();
         if (readToken(m->file).isWord("stream")) {
-            // The PDF specification states that the word "stream" should be followed by either a
+            readStream(object, og, offset);
-            // carriage return and a newline or by a newline alone.  It specifically disallowed
+        } else {
-            // following it by a carriage return alone since, in that case, there would be no way to
+            m->file->seek(cur_offset, SEEK_SET);
-            // tell whether the NL in a CR NL sequence was part of the stream data.  However, some
+        }
-            // readers, including Adobe reader, accept a carriage return by itself when followed by
+    }
-            // a non-newline character, so that's what we do here. We have also seen files that have
+
-            // extraneous whitespace between the stream keyword and the newline.
+    // Override last_offset so that it points to the beginning of the object we just read
+    m->file->setLastOffset(offset);
+    return object;
+}
+
+// After reading stream dictionary and stream keyword, read rest of stream.
+void
+QPDF::readStream(QPDFObjectHandle& object, QPDFObjGen og, qpdf_offset_t offset)
+{
+    // The PDF specification states that the word "stream" should be followed by either a carriage
+    // return and a newline or by a newline alone.  It specifically disallowed following it by a
+    // carriage return alone since, in that case, there would be no way to tell whether the NL in a
+    // CR NL sequence was part of the stream data.  However, some readers, including Adobe reader,
+    // accept a carriage return by itself when followed by a non-newline character, so that's what
+    // we do here. We have also seen files that have extraneous whitespace between the stream
+    // keyword and the newline.
     bool done = false;
     while (!done) {
         done = true;
         char ch;
         if (m->file->read(&ch, 1) == 0) {
-                    // A premature EOF here will result in some other problem that will get reported
+            // A premature EOF here will result in some other problem that will get reported at
-                    // at another time.
+            // another time.
         } else if (ch == '\n') {
             // ready to read stream data
             QTC::TC("qpdf", "QPDF stream with NL only");
@@ -1333,18 +1348,16 @@ QPDF::readObject(std::string const& description, QPDFObjGen og)
                     // Ready to read stream data
                     QTC::TC("qpdf", "QPDF stream with CRNL");
                 } else {
-                            // Treat the \r by itself as the whitespace after endstream and start
+                    // Treat the \r by itself as the whitespace after endstream and start reading
-                            // reading stream data in spite of not having seen a newline.
+                    // stream data in spite of not having seen a newline.
                     QTC::TC("qpdf", "QPDF stream with CR only");
                     m->file->unreadCh(ch);
                     warn(damagedPDF(
-                                m->file->tell(),
+                        m->file->tell(), "stream keyword followed by carriage return only"));
-                                "stream keyword followed by carriage return only"));
                 }
             }
         } else if (QUtil::is_space(ch)) {
-                    warn(damagedPDF(
+            warn(damagedPDF(m->file->tell(), "stream keyword followed by extraneous whitespace"));
-                        m->file->tell(), "stream keyword followed by extraneous whitespace"));
             done = false;
         } else {
             QTC::TC("qpdf", "QPDF stream without newline");
@@ -1388,14 +1401,6 @@ QPDF::readObject(std::string const& description, QPDFObjGen og)
         }
     }
     object = newIndirect(og, QPDF_Stream::create(this, og, object, stream_offset, length));
-        } else {
-            m->file->seek(cur_offset, SEEK_SET);
-        }
-    }
-
-    // Override last_offset so that it points to the beginning of the object we just read
-    m->file->setLastOffset(offset);
-    return object;
 }
 
 QPDFObjectHandle