simple clipping

share/base.css.in

@@ -49,7 +49,7 @@
   .@CSS_CSS_DRAW_CN@ { display:none; }
 }
 /* Part 2: Page Elements: Modify with caution
- * The followings are base classes, which are meant to be override by PDF specific classes
+ * The followings are base classes, some of which are meant to be override by PDF specific classes
  * So do not increase the specificity (e.g. ".classname" -> "#page-container .classname")
  */
 .@CSS_PAGE_DECORATION_CN@ { /* page decoration */
@@ -71,12 +71,15 @@
 .@CSS_PAGE_CONTENT_BOX_CN@ { /* content of a page */
   position:absolute;
   border-width:0;
+  padding:0;
+  margin:0;
   top:0;
   left:0;
   width:100%;
   height:100%;
   overflow:hidden;
   display:block;
+  /* set transform-origin for scaling */
   transform-origin:0% 0%;
   -ms-transform-origin:0% 0%;
   -moz-transform-origin:0% 0%;
@@ -114,6 +117,14 @@
     .@CSS_PAGE_CONTENT_BOX_CN@ {overflow:visible;}
   }
 }
+.@CSS_CLIP_CN@ { /* clip box */
+  position:absolute;
+  border-width:0;
+  padding:0;
+  margin:0;
+  overflow:hidden;
+  display:block;
+}
 .@CSS_LINE_CN@ { /* text line */
   position:absolute;
   white-space:pre;
@@ -144,7 +155,7 @@ span { /* text blocks within a line */
 .@CSS_PAGE_DATA_CN@ { /* info for Javascript */
   display:none;
 }
-.@CSS_LINE_CN@ { /* annotation links */
+.@CSS_LINK_CN@ { /* annotation links */
 }
 /* transparent color - WebKit */
 .@CSS_CSS_DRAW_CN@ { /* css drawing */

src/HTMLRenderer/HTMLRenderer.h

@@ -296,8 +296,7 @@ protected:
     {
         NLS_NONE,
         NLS_NEWSTATE, 
-        NLS_NEWLINE,
+        NLS_NEWLINE
-        NLS_NEWCLIP  
     } new_line_state;
     
     // for font reencoding

src/HTMLRenderer/general.cc

@@ -173,8 +173,13 @@ void HTMLRenderer::startPage(int pageNum, GfxState *state, XRef * xref)
 {
     this->pageNum = pageNum;
 
-    long long wid = all_manager.width.install(state->getPageWidth());
+    double pageWidth = state->getPageWidth();
-    long long hid = all_manager.height.install(state->getPageHeight());
+    double pageHeight = state->getPageHeight();
+
+    html_text_page.set_page_size(pageWidth, pageHeight);
+
+    long long wid = all_manager.width.install(pageWidth);
+    long long hid = all_manager.height.install(pageHeight);
     f_pages.fs 
         << "<div class=\"" << CSS::PAGE_DECORATION_CN 
             << " " << CSS::WIDTH_CN << wid

src/HTMLRenderer/state.cc

@@ -152,6 +152,13 @@ void HTMLRenderer::check_state_change(GfxState * state)
     
     new_line_state = NLS_NONE;
 
+    if(all_changed || clip_changed)
+    {
+        double x1, x2, y1, y2;
+        state->getClipBBox(&x1, &y1, &x2, &y2);
+        html_text_page.clip(x1, y1, x2, y2);
+    }
+
     bool need_recheck_position = false;
     bool need_rescale_font = false;
     bool draw_text_scale_changed = false;

src/HTMLTextLine.cc

@@ -29,6 +29,8 @@ HTMLTextLine::HTMLTextLine (const HTMLLineState & line_state, const Param & para
     :param(param)
     ,all_manager(all_manager) 
     ,line_state(line_state)
+    ,clip_x1(0)
+    ,clip_y1(0)
 { }
 
 void HTMLTextLine::append_unicodes(const Unicode * u, int l)
@@ -81,9 +83,9 @@ void HTMLTextLine::dump_text(ostream & out)
         // open <div> for the current text line
         out << "<div class=\"" << CSS::LINE_CN
             << " " << CSS::TRANSFORM_MATRIX_CN << all_manager.transform_matrix.install(line_state.transform_matrix)
-            << " " << CSS::LEFT_CN             << all_manager.left.install(line_state.x)
+            << " " << CSS::LEFT_CN             << all_manager.left.install(line_state.x - clip_x1)
             << " " << CSS::HEIGHT_CN           << all_manager.height.install(ascent)
-            << " " << CSS::BOTTOM_CN           << all_manager.bottom.install(line_state.y)
+            << " " << CSS::BOTTOM_CN           << all_manager.bottom.install(line_state.y - clip_y1)
             ;
         // it will be closed by the first state
     }
@@ -231,6 +233,12 @@ void HTMLTextLine::clear(void)
     text.clear();
 }
 
+void HTMLTextLine::clip(double x1, double y1, double x2, double y2)
+{
+    clip_x1 = x1;
+    clip_y1 = y1;
+}
+
 void HTMLTextLine::prepare(void)
 {
     if(param.optimize_text)

src/HTMLTextLine.h

@@ -81,6 +81,8 @@ public:
     bool text_empty(void) const { return text.empty(); }
     void clear(void);
 
+    void clip(double x1, double y1, double x2, double y2);
+
     /*
      * Optimize and calculate necessary values
      */
@@ -93,6 +95,7 @@ private:
 
     HTMLLineState line_state;
     double ascent, descent;
+    double clip_x1, clip_y1;
 
     std::vector<State> states;
     std::vector<Offset> offsets;

src/HTMLTextPage.cc

@@ -7,6 +7,7 @@
  */
 
 #include "HTMLTextPage.h"
+#include "util/css_const.h"
 
 namespace pdf2htmlEX {
 
@@ -17,6 +18,8 @@ HTMLTextPage::HTMLTextPage(const Param & param, AllStateManager & all_manager)
     : param(param)
     , all_manager(all_manager)
     , cur_line(nullptr)
+    , page_width(0)
+    , page_height(0)
 { } 
 
 void HTMLTextPage::dump_text(ostream & out)
@@ -25,8 +28,42 @@ void HTMLTextPage::dump_text(ostream & out)
         (*iter)->prepare();
     if(param.optimize_text)
         optimize();
-    for(auto iter = text_lines.begin(); iter != text_lines.end(); ++iter)
+
-        (*iter)->dump_text(out);
+    //push a dummy entry for convenience
+    clip_boxes.emplace_back(0, 0, page_width, page_height, text_lines.size());
+
+    ClipBox cur_cb(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)
+    {
+        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)
+        {
+            (*text_line_iter)->clip(cur_cb.x1, cur_cb.y1, cur_cb.x2, cur_cb.y2);
+            (*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) 
+                && equal(page_width, cur_cb.x2) && equal(page_height, cur_cb.y2));
+    }
 }
 
 void HTMLTextPage::dump_css(ostream & out)
@@ -37,6 +74,7 @@ void HTMLTextPage::dump_css(ostream & out)
 void HTMLTextPage::clear(void)
 {
     text_lines.clear();
+    clip_boxes.clear();
     cur_line = nullptr;
 }
 
@@ -50,6 +88,40 @@ void HTMLTextPage::open_new_line(const HTMLLineState & line_state)
     cur_line = text_lines.back().get();
 }
 
+void HTMLTextPage::set_page_size(double width, double height)
+{
+    page_width = width;
+    page_height = height;
+}
+
+void HTMLTextPage::clip(double x1, double y1, double x2, double y2)
+{
+    if(!clip_boxes.empty())
+    {
+        auto & cb = clip_boxes.back();
+        if(cb.start_idx == text_lines.size())
+        {
+            /*
+             * Previous ClipBox is not used
+             */
+            cb.x1 = x1;
+            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());
+}
+
 void HTMLTextPage::optimize(void)
 {
     //TODO

src/HTMLTextPage.h

@@ -35,6 +35,10 @@ public:
     void clear(void);
 
     void open_new_line(const HTMLLineState & line_state);
+    
+    /* for clipping */
+    void set_page_size(double width, double height);
+    void clip(double x1, double y1, double x2, double y2);
 
 private:
     void optimize(void);
@@ -42,7 +46,18 @@ private:
     const Param & param;
     AllStateManager & all_manager;
     HTMLTextLine * cur_line;
+    double page_width, page_height;
+
     std::vector<std::unique_ptr<HTMLTextLine>> text_lines;
+
+    struct ClipBox {
+        ClipBox(double x1, double y1, double x2, double y2, size_t start_idx)
+            :x1(x1),y1(y1),x2(x2),y2(y2),start_idx(start_idx)
+        { }
+        double x1, y1, x2, y2;
+        size_t start_idx;
+    };
+    std::vector<ClipBox> clip_boxes;
 };
 
 } //namespace pdf2htmlEX 

src/css_class_names.cmakelists.txt

@@ -9,6 +9,7 @@ set(CSS_INVALID_ID          "_")
 
 set(CSS_LINE_CN             "t") # Text 
 set(CSS_TRANSFORM_MATRIX_CN "m") # Matrix
+set(CSS_CLIP_CN             "c") # Clip
 
 set(CSS_PAGE_DECORATION_CN  "pd") # Page Decoration
 set(CSS_PAGE_FRAME_CN       "pf") # Page Frame

src/util/css_const.h.in

@@ -26,6 +26,7 @@ const char * const INVALID_ID          = "@CSS_INVALID_ID@";
 
 const char * const LINE_CN             = "@CSS_LINE_CN@";
 const char * const TRANSFORM_MATRIX_CN = "@CSS_TRANSFORM_MATRIX_CN@";
+const char * const CLIP_CN             = "@CSS_CLIP_CN@";
 
 // page_decoration is for shadow etc
 // page_frame cannot have margin or border-width, pdf2htmlEX.js will use it to determine the coordinates