Improve pdf-invert-images example

Jay Berkenbilt
1 parent fbac4725
Showing 2 changed files with 79 additions and 40 deletions
ChangeLog
examples/pdf-invert-images.cc
+2020-04-07  Jay Berkenbilt  <ejb@ql.org>
+
+	* Improve pdf-invert-images example to show a pattern of copying
+	streams into another QPDF object to enable a stream data provider
+	to access the original stream data.
+
+	* Fix error that caused a compilation error with clang. Fixes
+	#424.
+
 2020-04-06  Jay Berkenbilt  <ejb@ql.org>
 	* 10.0.0: release
@@ -21,28 +21,78 @@ void usage()
 }
 // Derive a class from StreamDataProvider to provide updated stream
-// data.  The main purpose of using this object is to avoid having to
-// allocate memory up front for the objects.  A real application might
-// use temporary files in order to avoid having to allocate all the
-// memory.  Here, we're not going to worry about that since the goal
-// is really to show how to use this facility rather than to show the
-// best possible way to write an image inverter.  This class still
-// illustrates dynamic creation of the new stream data.
+// data. The main purpose of using this object is to avoid having to
+// allocate memory up front for the objects. We want to replace the
+// stream data with a function of the original stream data. In order
+// to do this without actually holding all the images in memory, we
+// create another QPDF object and copy the streams. Copying the
+// streams doesn't actually copy the data. Internally, the qpdf
+// library is holding onto the location of the stream data in the
+// original file, which makes it possible for the StreamDataProvider
+// to access it when it needs it.
 class ImageInverter: public QPDFObjectHandle::StreamDataProvider
 {
   public:
+    ImageInverter();
     virtual ~ImageInverter()
     {
     }
     virtual void provideStreamData(int objid, int generation,
-				   Pipeline* pipeline);
+				   Pipeline* pipeline) override;
-    // Map [og] = image object
-    std::map<QPDFObjGen, QPDFObjectHandle> image_objects;
-    // Map [og] = image data
-    std::map<QPDFObjGen, PointerHolder<Buffer> > image_data;
+    void setSelfPh(PointerHolder<QPDFObjectHandle::StreamDataProvider>);
+    void registerImage(QPDFObjectHandle image);
+
+  private:
+    QPDF other;
+    PointerHolder<QPDFObjectHandle::StreamDataProvider> self_ph;
+    // Map og in original to copied image
+    std::map<QPDFObjGen, QPDFObjectHandle> copied_images;
 };
+ImageInverter::ImageInverter()
+{
+    this->other.emptyPDF();
+}
+
+void
+ImageInverter::setSelfPh(PointerHolder<QPDFObjectHandle::StreamDataProvider> p)
+{
+    // replaceStreamData requires a pointer holder to the stream data
+    // provider, but there's no way for us to generate one ourselves,
+    // so we have to have it handed to us.
+    this->self_ph = p;
+}
+
+
+void
+ImageInverter::registerImage(QPDFObjectHandle image)
+{
+    QPDFObjGen og(image.getObjGen());
+    // Store information about the images based on the object and
+    // generation number. Recall that a single image object may be
+    // used more than once, so no need to update the same stream
+    // multiple times.
+    if (this->copied_images.count(og) > 0)
+    {
+        return;
+    }
+    this->copied_images[og] = this->other.copyForeignObject(image);
+
+    // Register our stream data provider for this stream. Future calls
+    // to getStreamData or pipeStreamData will use the new
+    // information. Provide null for both filter and decode
+    // parameters. Note that this does not mean the image data will be
+    // uncompressed when we write the file. By default, QPDFWriter
+    // will use /FlateDecode for anything that is uncompressed or
+    // filterable in the input QPDF object, so we don't have to deal
+    // with it explicitly here. We could explicitly use /DCTDecode and
+    // write through a DCT filter if we wanted.
+    image.replaceStreamData(this->self_ph,
+			    QPDFObjectHandle::newNull(),
+			    QPDFObjectHandle::newNull());
+}
+
 void
 ImageInverter::provideStreamData(int objid, int generation,
 				 Pipeline* pipeline)
@@ -50,8 +100,9 @@ ImageInverter::provideStreamData(int objid, int generation,
     // Use the object and generation number supplied to look up the
     // image data.  Then invert the image data and write the inverted
     // data to the pipeline.
+    QPDFObjGen og(objid, generation);
     PointerHolder<Buffer> data =
-        this->image_data[QPDFObjGen(objid, generation)];
+        this->copied_images[og].getStreamData(qpdf_dl_all);
     size_t size = data->getSize();
     unsigned char* buf = data->getBuffer();
     unsigned char ch;
@@ -98,6 +149,9 @@ int main(int argc, char* argv[])
 	ImageInverter* inv = new ImageInverter;
 	PointerHolder<QPDFObjectHandle::StreamDataProvider> p = inv;
+        // We need to give ImageInverter the pointer holder that it
+        // needs to pass to replaceStreamData.
+        inv->setSelfPh(p);
 	// For each page...
 	std::vector<QPDFPageObjectHelper> pages =
@@ -126,38 +180,14 @@ int main(int argc, char* argv[])
 		// whether the image is filterable.  Directly inspect
 		// keys to determine the image type.
 		if (image.pipeStreamData(0, qpdf_ef_compress,
-                                         qpdf_dl_generalized) &&
+                                         qpdf_dl_all) &&
 		    color_space.isName() &&
 		    bits_per_component.isInteger() &&
 		    (color_space.getName() == "/DeviceGray") &&
 		    (bits_per_component.getIntValue() == 8))
 		{
-		    // Store information about the images based on the
-		    // object and generation number.  Recall that a single
-		    // image object may be used more than once.
-		    QPDFObjGen og = image.getObjGen();
-		    if (inv->image_objects.count(og) == 0)
-		    {
-			inv->image_objects[og] = image;
-			inv->image_data[og] = image.getStreamData();
-
-			// Register our stream data provider for this
-			// stream.  Future calls to getStreamData or
-			// pipeStreamData will use the new
-			// information.  Provide null for both filter
-			// and decode parameters.  Note that this does
-			// not mean the image data will be
-			// uncompressed when we write the file.  By
-			// default, QPDFWriter will use /FlateDecode
-			// for anything that is uncompressed or
-			// filterable in the input QPDF object, so we
-			// don't have to deal with it explicitly here.
-			image.replaceStreamData(
-			    p,
-			    QPDFObjectHandle::newNull(),
-			    QPDFObjectHandle::newNull());
-		    }
-		}
+                    inv->registerImage(image);
+                }
 	    }
 	}