do not embed empty bg image

src/BackgroundRenderer/SplashBackgroundRenderer.cc

@@ -4,16 +4,32 @@
  * Copyright (C) 2012 Lu Wang <coolwanglu@gmail.com>
  */
 
+#include <fstream>
+
 #include <PDFDoc.h>
 
+#include "Base64Stream.h"
+
 #include "SplashBackgroundRenderer.h"
 
 namespace pdf2htmlEX {
 
 using std::string;
+using std::ifstream;
 
 const SplashColor SplashBackgroundRenderer::white = {255,255,255};
 
+/*
+ * SplashOutputDev::startPage would paint the whole page with the background color
+ * And thus have modified region set to the whole page area
+ * We do not want that.
+ */
+void SplashBackgroundRenderer::startPage(int pageNum, GfxState *state, XRef *xrefA)
+{
+    SplashOutputDev::startPage(pageNum, state, xrefA);
+    clearModRegion();
+}
+
 void SplashBackgroundRenderer::drawChar(GfxState *state, double x, double y,
   double dx, double dy,
   double originX, double originY,
@@ -40,17 +56,50 @@ static GBool annot_cb(Annot *, void *) {
     return false;
 };
 
-void SplashBackgroundRenderer::render_page(PDFDoc * doc, int pageno, const string & filename)
+void SplashBackgroundRenderer::render_page(PDFDoc * doc, int pageno)
 {
     doc->displayPage(this, pageno, param.h_dpi, param.v_dpi,
             0, 
             (!(param.use_cropbox)),
             false, false,
             nullptr, nullptr, &annot_cb, nullptr);
+}
+
+void SplashBackgroundRenderer::embed_image(int pageno)
+{
+    int xmin, xmax, ymin, ymax;
+    getModRegion(&xmin, &ymin, &xmax, &ymax);
+
+    // dump the background image only when it is not empty
+    if((xmin <= xmax) && (ymin <= ymax))
+    {
+        {
+            auto fn = html_renderer->str_fmt("%s/bg%x.png", (param.embed_image ? param.tmp_dir : param.dest_dir).c_str(), pageno);
+            if(param.embed_image)
+                html_renderer->tmp_files.add((char*)fn);
 
             getBitmap()->writeImgFile(splashFormatPng, 
-            (char*)filename.c_str(),
+                    (char*)fn,
                     param.h_dpi, param.v_dpi);
         }
 
+        auto & f_page = *(html_renderer->f_curpage);
+        f_page << "<img class=\"" << CSS::BACKGROUND_IMAGE_CN 
+            << "\" alt=\"\" src=\"";
+        if(param.embed_image)
+        {
+            auto path = html_renderer->str_fmt("%s/bg%x.png", param.tmp_dir.c_str(), pageno);
+            ifstream fin((char*)path, ifstream::binary);
+            if(!fin)
+                throw string("Cannot read background image ") + (char*)path;
+            f_page << "data:image/png;base64," << Base64Stream(fin);
+        }
+        else
+        {
+            f_page << (char*)html_renderer->str_fmt("bg%x.png", pageno);
+        }
+        f_page << "\"/>";
+    }
+}
+
 } // namespace pdf2htmlEX

src/BackgroundRenderer/SplashBackgroundRenderer.h

@@ -27,13 +27,15 @@ public:
   static const SplashColor white;
 
   SplashBackgroundRenderer(HTMLRenderer * html_renderer, const Param & param)
-      : SplashOutputDev(splashModeRGB8, 4, gFalse, (SplashColorPtr)&white, gTrue, gTrue)
+      : SplashOutputDev(splashModeRGB8, 4, gFalse, (SplashColorPtr)(&white), gTrue, gTrue)
       , html_renderer(html_renderer)
       , param(param)
   { }
 
   virtual ~SplashBackgroundRenderer() { }
 
+  virtual void startPage(int pageNum, GfxState *state, XRef *xrefA);
+  
   virtual void drawChar(GfxState *state, double x, double y,
       double dx, double dy,
       double originX, double originY,
@@ -49,7 +51,8 @@ public:
           SplashOutputDev::fill(state);
   }
 
-  void render_page(PDFDoc * doc, int pageno, const std::string & filename);
+  void render_page(PDFDoc * doc, int pageno);
+  void embed_image(int pageno);
 
 protected:
   HTMLRenderer * html_renderer;

src/HTMLRenderer/HTMLRenderer.h

@@ -28,6 +28,8 @@
 #include "StateManager.h"
 #include "HTMLTextPage.h"
 
+#include "BackgroundRenderer/BackgroundRenderer_forward.h"
+
 #include "util/const.h"
 #include "util/misc.h"
 
@@ -307,11 +309,18 @@ protected:
     int * width_list;
 
     Preprocessor preprocessor;
+
+    // manage temporary files
     TmpFiles tmp_files;
 
     // for string formatting
     StringFormatter str_fmt;
 
+    // render background image
+    friend BackgroundRenderer;
+    BackgroundRenderer * bg_renderer;
+
+
     struct {
         std::ofstream fs;
         std::string path;

src/HTMLRenderer/general.cc

@@ -17,9 +17,10 @@
 #include "pdf2htmlEX-config.h"
 #include "HTMLRenderer.h"
 #include "HTMLTextLine.h"
-#include "BackgroundRenderer/BackgroundRenderer.h"
 #include "Base64Stream.h"
 
+#include "BackgroundRenderer/BackgroundRenderer.h"
+
 #include "util/namespace.h"
 #include "util/ffw.h"
 #include "util/math.h"
@@ -96,7 +97,7 @@ void HTMLRenderer::process(PDFDoc *doc)
     ///////////////////
     // Process pages
     
-    BackgroundRenderer * bg_renderer = nullptr;
+    bg_renderer = nullptr;
     if(param.process_nontext)
     {
         bg_renderer = new BackgroundRenderer(this, param);
@@ -122,11 +123,7 @@ void HTMLRenderer::process(PDFDoc *doc)
 
         if(param.process_nontext)
         {
-            auto fn = str_fmt("%s/bg%x.png", (param.embed_image ? param.tmp_dir : param.dest_dir).c_str(), i);
-            if(param.embed_image)
-                tmp_files.add((char*)fn);
-
-            bg_renderer->render_page(doc, i, (char*)fn);
+            bg_renderer->render_page(doc, i);
         }
 
         doc->displayPage(this, i, 
@@ -155,7 +152,10 @@ void HTMLRenderer::process(PDFDoc *doc)
     post_process();
 
     if(bg_renderer)
+    {
         delete bg_renderer;
+        bg_renderer = nullptr;
+    }
 
     cerr << endl;
 }
@@ -215,21 +215,7 @@ void HTMLRenderer::startPage(int pageNum, GfxState *state, XRef * xref)
 
     if(param.process_nontext)
     {
-        (*f_curpage) << "<img class=\"" << CSS::BACKGROUND_IMAGE_CN 
-            << "\" alt=\"\" src=\"";
-        if(param.embed_image)
-        {
-            auto path = str_fmt("%s/bg%x.png", param.tmp_dir.c_str(), pageNum);
-            ifstream fin((char*)path, ifstream::binary);
-            if(!fin)
-                throw string("Cannot read background image ") + (char*)path;
-            (*f_curpage) << "data:image/png;base64," << Base64Stream(fin);
-        }
-        else
-        {
-            (*f_curpage) << (char*)str_fmt("bg%x.png", pageNum);
-        }
-        (*f_curpage) << "\"/>";
+        bg_renderer->embed_image(pageNum);
     }
 
     reset_state();