check clip changes

src/HTMLRenderer/HTMLRenderer.h

@@ -289,6 +289,7 @@ protected:
     AllStateManager all_manager;
     HTMLTextState cur_text_state;
     HTMLLineState cur_line_state;
+    HTMLClipState cur_clip_state;
 
     HTMLTextPage html_text_page;
 
@@ -296,7 +297,8 @@ protected:
     {
         NLS_NONE,
         NLS_NEWSTATE, 
-        NLS_NEWLINE
+        NLS_NEWLINE,
+        NLS_NEWCLIP
     } new_line_state;
     
     // for font reencoding

src/HTMLRenderer/state.cc

@@ -119,6 +119,11 @@ void HTMLRenderer::reset_state()
     cur_line_state.y = 0;
     memcpy(cur_line_state.transform_matrix, ID_MATRIX, sizeof(cur_line_state.transform_matrix));
 
+    cur_clip_state.xmin = 0;
+    cur_clip_state.xmax = 0;
+    cur_clip_state.ymin = 0;
+    cur_clip_state.ymax = 0;
+
     cur_tx  = cur_ty  = 0;
     draw_tx = draw_ty = 0;
 
@@ -145,6 +150,14 @@ void HTMLRenderer::reset_state_change()
 
     clip_changed = false;
 }
+
+template<class NewLineState>
+void set_line_state(NewLineState & cur_ls, NewLineState new_ls)
+{
+    if(new_ls > cur_ls)
+        cur_ls = new_ls;
+}
+
 void HTMLRenderer::check_state_change(GfxState * state)
 {
     // DEPENDENCY WARNING
@@ -154,11 +167,16 @@ void HTMLRenderer::check_state_change(GfxState * state)
 
     if(all_changed || clip_changed)
     {
-        //TODO: compare with current clip box
+        HTMLClipState new_clip_state;
-        double x1, x2, y1, y2;
+        state->getClipBBox(&new_clip_state.xmin, &new_clip_state.ymin, &new_clip_state.xmax, &new_clip_state.ymax);
-        state->getClipBBox(&x1, &y1, &x2, &y2);
+        if(!(equal(cur_clip_state.xmin, new_clip_state.xmin)
-        html_text_page.clip(x1, y1, x2, y2);
+                    && equal(cur_clip_state.xmax, new_clip_state.xmax)
-        new_line_state = NLS_NEWLINE;
+                    && equal(cur_clip_state.ymin, new_clip_state.ymin)
+                    && equal(cur_clip_state.ymax, new_clip_state.ymax)))
+        {
+            cur_clip_state = new_clip_state;
+            set_line_state(new_line_state, NLS_NEWCLIP);
+        }
     }
 
     bool need_recheck_position = false;
@@ -191,11 +209,11 @@ void HTMLRenderer::check_state_change(GfxState * state)
             // TODO: consider the font matrix and estimate the metrics  
             if(new_font_info->is_type3 || cur_text_state.font_info->is_type3)
             {
-                new_line_state = max<NewLineState>(new_line_state, NLS_NEWLINE);
+                set_line_state(new_line_state, NLS_NEWLINE);
             }
             else
             {
-                new_line_state = max<NewLineState>(new_line_state, NLS_NEWSTATE);
+                set_line_state(new_line_state, NLS_NEWSTATE);
             }
             cur_text_state.font_info = new_font_info;
         }
@@ -279,13 +297,13 @@ void HTMLRenderer::check_state_change(GfxState * state)
 
         if(!equal(new_draw_font_size, cur_text_state.font_size))
         {
-            new_line_state = max<NewLineState>(new_line_state, NLS_NEWSTATE);
+            set_line_state(new_line_state, NLS_NEWSTATE);
             cur_text_state.font_size = new_draw_font_size;
         }
 
         if(!tm_equal(new_draw_text_tm, cur_line_state.transform_matrix, 4))
         {
-            new_line_state = max<NewLineState>(new_line_state, NLS_NEWLINE);
+            set_line_state(new_line_state, NLS_NEWLINE);
             memcpy(cur_line_state.transform_matrix, new_draw_text_tm, sizeof(cur_line_state.transform_matrix));
         }
     }
@@ -368,14 +386,14 @@ void HTMLRenderer::check_state_change(GfxState * state)
             else
             {
                 cur_text_state.vertical_align = (dy * old_draw_text_scale);
-                new_line_state = max<NewLineState>(new_line_state, NLS_NEWSTATE);
+                set_line_state(new_line_state, NLS_NEWSTATE);
             }
             draw_tx = cur_tx;
             draw_ty = cur_ty;
         }
         else
         {
-            new_line_state = max<NewLineState>(new_line_state, NLS_NEWLINE);
+            set_line_state(new_line_state, NLS_NEWLINE);
         }
     }
     else
@@ -392,7 +410,7 @@ void HTMLRenderer::check_state_change(GfxState * state)
         if(!equal(new_letter_space, cur_text_state.letter_space))
         {
             cur_text_state.letter_space = new_letter_space;
-            new_line_state = max<NewLineState>(new_line_state, NLS_NEWSTATE);
+            set_line_state(new_line_state, NLS_NEWSTATE);
         }
     }
 
@@ -404,7 +422,7 @@ void HTMLRenderer::check_state_change(GfxState * state)
         if(!equal(new_word_space, cur_text_state.word_space))
         {
             cur_text_state.word_space = new_word_space;
-            new_line_state = max<NewLineState>(new_line_state, NLS_NEWSTATE);
+            set_line_state(new_line_state, NLS_NEWSTATE);
         }
     }
 
@@ -429,7 +447,7 @@ void HTMLRenderer::check_state_change(GfxState * state)
         if(!(new_fill_color == cur_text_state.fill_color))
         {
             cur_text_state.fill_color = new_fill_color;
-            new_line_state = max<NewLineState>(new_line_state, NLS_NEWSTATE);
+            set_line_state(new_line_state, NLS_NEWSTATE);
         }
     }
 
@@ -455,7 +473,7 @@ void HTMLRenderer::check_state_change(GfxState * state)
         if(!(new_stroke_color == cur_text_state.stroke_color))
         {
             cur_text_state.stroke_color = new_stroke_color;
-            new_line_state = max<NewLineState>(new_line_state, NLS_NEWSTATE);
+            set_line_state(new_line_state, NLS_NEWSTATE);
         }
     }
 
@@ -465,9 +483,14 @@ void HTMLRenderer::check_state_change(GfxState * state)
 void HTMLRenderer::prepare_text_line(GfxState * state)
 {
     if(!(html_text_page.get_cur_line()))
-        new_line_state = NLS_NEWLINE;
+        new_line_state = NLS_NEWCLIP;
 
-    if(new_line_state == NLS_NEWLINE)
+    if(new_line_state >= NLS_NEWCLIP)
+    {
+        html_text_page.clip(cur_clip_state);
+    }
+    
+    if(new_line_state >= NLS_NEWLINE)
     {
         // update position such that they will be recorded by text_line_buf
         state->transform(state->getCurX(), state->getCurY(), &cur_line_state.x, &cur_line_state.y);

src/HTMLTextLine.cc

@@ -233,10 +233,10 @@ void HTMLTextLine::clear(void)
     text.clear();
 }
 
-void HTMLTextLine::clip(double x1, double y1, double x2, double y2)
+void HTMLTextLine::clip(const HTMLClipState & clip_state)
 {
-    clip_x1 = x1;
+    clip_x1 = clip_state.xmin;
-    clip_y1 = y1;
+    clip_y1 = clip_state.ymin;
 }
 
 void HTMLTextLine::prepare(void)

src/HTMLTextLine.h

@@ -81,7 +81,7 @@ public:
     bool text_empty(void) const { return text.empty(); }
     void clear(void);
 
-    void clip(double x1, double y1, double x2, double y2);
+    void clip(const HTMLClipState &);
 
     /*
      * Optimize and calculate necessary values

src/HTMLTextPage.cc

@@ -30,39 +30,49 @@ void HTMLTextPage::dump_text(ostream & out)
         optimize();
 
     //push a dummy entry for convenience
-    clip_boxes.emplace_back(0, 0, page_width, page_height, text_lines.size());
+    clips.emplace_back(HTMLClipState{0, 0, page_width, page_height}, text_lines.size());
 
-    ClipBox cur_cb(0, 0, page_width, page_height, 0);
+    Clip cur_clip(HTMLClipState{0, 0, page_width, page_height}, 0);
     bool has_clip = false;
 
     auto text_line_iter = text_lines.begin();
-    for(auto clip_iter = clip_boxes.begin(); clip_iter != clip_boxes.end(); ++clip_iter)
+    for(auto clip_iter = clips.begin(); clip_iter != clips.end(); ++clip_iter)
     {
-        if(has_clip)
-        {
-            out << "<div class=\"" << CSS::CLIP_CN
-                << " " << CSS::LEFT_CN   << all_manager.left.install(cur_cb.x1)
-                << " " << CSS::BOTTOM_CN << all_manager.bottom.install(cur_cb.y1)
-                << " " << CSS::WIDTH_CN  << all_manager.width.install(cur_cb.x2 - cur_cb.x1)
-                << " " << CSS::HEIGHT_CN << all_manager.height.install(cur_cb.y2 - cur_cb.y1)
-                << "\">";
-        }
-
         auto next_text_line_iter = text_lines.begin() + clip_iter->start_idx;
-        while(text_line_iter != next_text_line_iter)
+        if(text_line_iter != next_text_line_iter)
         {
-            (*text_line_iter)->clip(cur_cb.x1, cur_cb.y1, cur_cb.x2, cur_cb.y2);
+            const auto & cs = cur_clip.clip_state;
-            (*text_line_iter)->dump_text(out);
+            if(has_clip)
-            ++text_line_iter;
+            {
-        }
+                out << "<div class=\"" << CSS::CLIP_CN
-        if(has_clip)
+                    << " " << CSS::LEFT_CN   << all_manager.left.install(cs.xmin)
-        {
+                    << " " << CSS::BOTTOM_CN << all_manager.bottom.install(cs.ymin)
-            out << "</div>";
+                    << " " << CSS::WIDTH_CN  << all_manager.width.install(cs.xmax - cs.xmin)
+                    << " " << CSS::HEIGHT_CN << all_manager.height.install(cs.ymax - cs.ymin)
+                    << "\">";
+            }
+
+            while(text_line_iter != next_text_line_iter)
+            {
+                if(has_clip)
+                {
+                    (*text_line_iter)->clip(cs);
+                }
+                (*text_line_iter)->dump_text(out);
+                ++text_line_iter;
+            }
+            if(has_clip)
+            {
+                out << "</div>";
+            }
         }
 
-        cur_cb = *clip_iter;
+        {
-        has_clip = !(equal(0, cur_cb.x1) && equal(0, cur_cb.y1) 
+            cur_clip = *clip_iter;
-                && equal(page_width, cur_cb.x2) && equal(page_height, cur_cb.y2));
+            const auto & cs = cur_clip.clip_state;
+            has_clip = !(equal(0, cs.xmin) && equal(0, cs.ymin)
+                    && equal(page_width, cs.xmax) && equal(page_height, cs.ymax));
+        }
     }
 }
 
@@ -74,7 +84,7 @@ void HTMLTextPage::dump_css(ostream & out)
 void HTMLTextPage::clear(void)
 {
     text_lines.clear();
-    clip_boxes.clear();
+    clips.clear();
     cur_line = nullptr;
 }
 
@@ -94,32 +104,21 @@ void HTMLTextPage::set_page_size(double width, double height)
     page_height = height;
 }
 
-void HTMLTextPage::clip(double x1, double y1, double x2, double y2)
+void HTMLTextPage::clip(const HTMLClipState & clip_state)
 {
-    if(!clip_boxes.empty())
+    if(!clips.empty())
     {
-        auto & cb = clip_boxes.back();
+        auto & clip = clips.back();
-        if(cb.start_idx == text_lines.size())
+        if(clip.start_idx == text_lines.size())
         {
             /*
              * Previous ClipBox is not used
              */
-            cb.x1 = x1;
+            clip.clip_state = clip_state;
-            cb.y1 = y1;
-            cb.x2 = x2;
-            cb.y2 = y2;
-            return;
-        }
-        if(equal(cb.x1, x1) && equal(cb.y1, y1)
-                && equal(cb.x2, x2) && equal(cb.y2, y2))
-        {
-            /*
-             * same as previous ClipBox
-             */
             return;
         }
     }
-    clip_boxes.emplace_back(x1, y1, x2, y2, text_lines.size());
+    clips.emplace_back(clip_state, text_lines.size());
 }
 
 void HTMLTextPage::optimize(void)

src/HTMLTextPage.h

@@ -22,7 +22,6 @@ namespace pdf2htmlEX {
  *
  * contains a series of HTMLTextLine
  */
-
 class HTMLTextPage
 {
 public:
@@ -38,7 +37,7 @@ public:
     
     /* for clipping */
     void set_page_size(double width, double height);
-    void clip(double x1, double y1, double x2, double y2);
+    void clip(const HTMLClipState & clip_state);
 
 private:
     void optimize(void);
@@ -50,14 +49,14 @@ private:
 
     std::vector<std::unique_ptr<HTMLTextLine>> text_lines;
 
-    struct ClipBox {
+    struct Clip {
-        ClipBox(double x1, double y1, double x2, double y2, size_t start_idx)
+        HTMLClipState clip_state;
-            :x1(x1),y1(y1),x2(x2),y2(y2),start_idx(start_idx)
-        { }
-        double x1, y1, x2, y2;
         size_t start_idx;
+        Clip(const HTMLClipState & clip_state, size_t start_idx)
+            :clip_state(clip_state),start_idx(start_idx)
+        { }
     };
-    std::vector<ClipBox> clip_boxes;
+    std::vector<Clip> clips;
 };
 
 } //namespace pdf2htmlEX