Merge pull request #371 from duanyao/proof_output

Implement proof output

pdf2htmlEX.1.in

@@ -296,6 +296,12 @@ Experimental and unsupported CSS drawing
 .B --debug <0|1> (Default: 0)
 Print debug information.
 
+.TP
+.B --proof <0|1|2> (Default: 0)
+Output a proof version. If a positive value is specified, texts are drawn on both text layer and background image for comparision.
+If 2 is specified, texts on background are in different colors. If png/jpg background format is used,
+a higher hdpi/vdpi (e.g. 288) is recommended for legibility.
+
 .SS Meta
 
 .TP

src/BackgroundRenderer/BackgroundRenderer.cc

@@ -49,4 +49,80 @@ BackgroundRenderer * BackgroundRenderer::getFallbackBackgroundRenderer(HTMLRende
     return nullptr;
 }
 
+void BackgroundRenderer::proof_begin_text_object(GfxState *state, OutputDev * dev)
+{
+    if (!proof_state)
+    {
+        PDFRectangle rect(0, 0, state->getPageWidth(), state->getPageHeight());
+        proof_state.reset(new GfxState(state->getHDPI(), state->getVDPI(), &rect, state->getRotate(), dev->upsideDown()));
+        proof_state->setFillColorSpace(new GfxDeviceRGBColorSpace());
+        proof_state->setStrokeColorSpace(new GfxDeviceRGBColorSpace());
+    }
+
+    int render = state->getRender();
+
+    // Save original render mode in proof_state, and restore in proof_end_text_object()
+    // This is due to poppler's OutputDev::updateRender() actually has no effect, we have to
+    // modify state directly, see proof_begin_string().
+    proof_state->setRender(render);
+}
+
+void BackgroundRenderer::proof_begin_string(GfxState *state, OutputDev * dev)
+{
+    int render = proof_state->getRender();
+    if (render == 3 || state->getRender() == 3) // hidden
+        return;
+
+    double lx = state->getFontSize() / 50, ly = lx;
+    tm_transform(state->getTextMat(), lx, ly, true);
+    proof_state->setLineWidth(std::min(fabs(lx), fabs(ly)));
+
+    static const Color red(1, 0, 0), green(0, 1, 0), blue(0, 0, 1), yellow(1, 1, 0), white(1, 1, 1);
+    Color fc, sc;
+    const Color *pfc, *psc;
+    state->getFillRGB(&fc.rgb);
+    state->getStrokeRGB(&sc.rgb);
+
+    if (render == 0 || render == 2) //has fill
+        pfc = fc.distance(red) >  0.4 ? &red : &green;
+    else
+        pfc = &red;
+
+    if (render == 1 || render == 2) // has stroke
+        psc = sc.distance(blue) >  0.4 ?  &blue : &yellow;
+    else if(render == 0) // fill only
+        psc = &white;
+    else
+        psc = &blue;
+
+    GfxColor gfc, gsc;
+    pfc->get_gfx_color(gfc);
+    psc->get_gfx_color(gsc);
+    proof_state->setFillColor(&gfc);
+    proof_state->setStrokeColor(&gsc);
+
+    if (state->getFillColorSpace()->getMode() != csDeviceRGB)
+        dev->updateFillColorSpace(proof_state.get());
+    if (state->getStrokeColorSpace()->getMode() != csDeviceRGB)
+        dev->updateStrokeColorSpace(proof_state.get());
+
+    dev->updateLineWidth(proof_state.get());
+    dev->updateFillColor(proof_state.get());
+    dev->updateStrokeColor(proof_state.get());
+
+    state->setRender(2); // fill & stroke
+    dev->updateRender(state);
+}
+
+void BackgroundRenderer::proof_end_text_object(GfxState *state, OutputDev * dev)
+{
+    state->setRender(proof_state->getRender());
+    dev->updateRender(state);
+    dev->updateLineWidth(state);
+    dev->updateFillColorSpace(state);
+    dev->updateStrokeColorSpace(state);
+    dev->updateFillColor(state);
+    dev->updateStrokeColor(state);
+}
+
 } // namespace pdf2htmlEX

src/BackgroundRenderer/BackgroundRenderer.h

@@ -10,8 +10,11 @@
 #define BACKGROUND_RENDERER_H__
 
 #include <string>
+#include <memory>
 
 class PDFDoc;
+class GfxState;
+class OutputDev;
 
 namespace pdf2htmlEX {
 
@@ -34,6 +37,13 @@ public:
     virtual bool render_page(PDFDoc * doc, int pageno) = 0;
     virtual void embed_image(int pageno) = 0;
 
+    // for proof output
+protected:
+    void proof_begin_text_object(GfxState * state, OutputDev * dev);
+    void proof_begin_string(GfxState * state, OutputDev * dev);
+    void proof_end_text_object(GfxState * state, OutputDev * dev);
+private:
+    std::unique_ptr<GfxState> proof_state;
 };
 
 } // namespace pdf2htmlEX

src/BackgroundRenderer/CairoBackgroundRenderer.cc

@@ -53,7 +53,7 @@ void CairoBackgroundRenderer::drawChar(GfxState *state, double x, double y,
     // - OR there is special filling method
     // - OR using a writing mode font
     // - OR using a Type 3 font while param.process_type3 is not enabled
-    if((param.fallback)
+    if((param.fallback || param.proof)
         || ( (state->getFont())
             && ( (state->getFont()->getWMode())
                  || ((state->getFont()->getType() == fontType3) && (!param.process_type3))
@@ -65,6 +65,27 @@ void CairoBackgroundRenderer::drawChar(GfxState *state, double x, double y,
     }
 }
 
+void CairoBackgroundRenderer::beginTextObject(GfxState *state)
+{
+    if (param.proof == 2)
+        proof_begin_text_object(state, this);
+    CairoOutputDev::beginTextObject(state);
+}
+
+void CairoBackgroundRenderer::beginString(GfxState *state, GooString * str)
+{
+    if (param.proof == 2)
+        proof_begin_string(state, this);
+    CairoOutputDev::beginString(state, str);
+}
+
+void CairoBackgroundRenderer::endTextObject(GfxState *state)
+{
+    if (param.proof == 2)
+        proof_end_text_object(state, this);
+    CairoOutputDev::endTextObject(state);
+}
+
 void CairoBackgroundRenderer::init(PDFDoc * doc)
 {
     startDoc(doc);

src/BackgroundRenderer/CairoBackgroundRenderer.h

@@ -44,6 +44,11 @@ public:
       double originX, double originY,
       CharCode code, int nBytes, Unicode *u, int uLen);
 
+  //for proof
+  void beginTextObject(GfxState *state);
+  void beginString(GfxState *state, GooString * str);
+  void endTextObject(GfxState *state);
+
 protected:
   virtual void setMimeData(Stream *str, Object *ref, cairo_surface_t *image);
 

src/BackgroundRenderer/SplashBackgroundRenderer.cc

@@ -78,7 +78,7 @@ void SplashBackgroundRenderer::drawChar(GfxState *state, double x, double y,
     // - OR there is special filling method
     // - OR using a writing mode font
     // - OR using a Type 3 font while param.process_type3 is not enabled
-    if((param.fallback)
+    if((param.fallback || param.proof)
        || ( (state->getFont()) 
             && ( (state->getFont()->getWMode())
                  || ((state->getFont()->getType() == fontType3) && (!param.process_type3))
@@ -90,6 +90,27 @@ void SplashBackgroundRenderer::drawChar(GfxState *state, double x, double y,
     }
 }
 
+void SplashBackgroundRenderer::beginTextObject(GfxState *state)
+{
+    if (param.proof == 2)
+        proof_begin_text_object(state, this);
+    SplashOutputDev::beginTextObject(state);
+}
+
+void SplashBackgroundRenderer::beginString(GfxState *state, GooString * str)
+{
+    if (param.proof == 2)
+        proof_begin_string(state, this);
+    SplashOutputDev::beginString(state, str);
+}
+
+void SplashBackgroundRenderer::endTextObject(GfxState *state)
+{
+    if (param.proof == 2)
+        proof_end_text_object(state, this);
+    SplashOutputDev::endTextObject(state);
+}
+
 void SplashBackgroundRenderer::init(PDFDoc * doc)
 {
     startDoc(doc);

src/BackgroundRenderer/SplashBackgroundRenderer.h

@@ -60,6 +60,11 @@ public:
           SplashOutputDev::fill(state);
   }
 
+  //for proof
+  void beginTextObject(GfxState *state);
+  void beginString(GfxState *state, GooString * str);
+  void endTextObject(GfxState *state);
+
 protected:
   void dump_image(const char * filename, int x1, int y1, int x2, int y2);
   HTMLRenderer * html_renderer;

src/Color.cc

@@ -1,3 +1,5 @@
+#include <cmath>
+
 #include "Color.h"
 
 #include "util/misc.h"
@@ -6,6 +8,22 @@ namespace pdf2htmlEX {
 
 using std::ostream;
 
+Color::Color()
+{
+    memset(this, 0, sizeof(Color));
+}
+
+Color::Color(double r, double g, double b, bool transparent)
+    :transparent(transparent)
+{
+    rgb.r = (GfxColorComp)(r * gfxColorComp1);
+    rgb.g = (GfxColorComp)(g * gfxColorComp1);
+    rgb.b = (GfxColorComp)(b * gfxColorComp1);
+}
+
+Color::Color(const GfxRGB& rgb)
+    :transparent(false), rgb(rgb) { }
+
 ostream & operator << (ostream & out, const Color & color)
 {
     if(color.transparent)
@@ -15,4 +33,19 @@ ostream & operator << (ostream & out, const Color & color)
     return out;
 }
 
+void Color::get_gfx_color(GfxColor & gc) const
+{
+    gc.c[0] = rgb.r;
+    gc.c[1] = rgb.g;
+    gc.c[2] = rgb.b;
+}
+
+double Color::distance(const Color & other) const
+{
+    double dr = (double)rgb.r - other.rgb.r,
+            dg = (double)rgb.g - other.rgb.g,
+            db = (double)rgb.b - other.rgb.b;
+    return sqrt((dr * dr + dg * dg + db * db) / (3.0 * gfxColorComp1 * gfxColorComp1));
+}
+
 } // namespace pdf2htmlEX

src/Color.h

@@ -16,6 +16,9 @@ struct Color
 {
     bool transparent;
     GfxRGB rgb;
+    Color();
+    Color(double r, double g, double b, bool transparent = false);
+    Color(const GfxRGB& rgb);
     bool operator == (const Color & c) const {
         if(transparent != c.transparent)
             return false;
@@ -23,6 +26,9 @@ struct Color
             return true;
         return ((rgb.r == c.rgb.r) && (rgb.g == c.rgb.g) && (rgb.b == c.rgb.b));
     }
+    void get_gfx_color(GfxColor & gc) const;
+    // Color distance, [0,1].
+    double distance(const Color & other) const;
 };
 
 std::ostream & operator << (std::ostream & out, const Color & color);

src/Param.h

@@ -76,6 +76,7 @@ struct Param
     std::string tmp_dir;
     int css_draw;
     int debug;
+    int proof;
 
     std::string input_filename, output_filename;
 };

src/pdf2htmlEX.cc

@@ -206,6 +206,7 @@ void parse_options (int argc, char **argv)
         // TODO: css drawings are hidden on print, for annot links, need to fix it for other drawings
 //        .add("css-draw", &param.css_draw, 0, "[experimental and unsupported] CSS drawing")
         .add("debug", &param.debug, 0, "print debugging information")
+        .add("proof", &param.proof, 0, "texts are drawn on both text layer and background for proof.")
 
         // meta
         .add("version,v", "print copyright and version info", &show_version_and_exit)