single js

CMakeLists.txt

@@ -136,6 +136,11 @@ add_executable(pdf2htmlEX
 target_link_libraries(pdf2htmlEX ${PDF2HTMLEX_LIBS})
 
 install (TARGETS pdf2htmlEX DESTINATION bin)
-file (GLOB datafiles share/*)
+file (GLOB datafiles 
+    share/base.css
+    share/jquery.js
+    share/pdf2htmlEX.js
+    share/manifest
+    )
 install (FILES ${datafiles} DESTINATION share/pdf2htmlEX)
 install (FILES pdf2htmlEX.1 DESTINATION share/man/man1)

share/base.css

@@ -1,8 +1,5 @@
 /* Base CSS */
 /* by Wang Lu */
-body {
-  position:relative;
-}
 #pdf-main {
   position:absolute;
   top:0;
@@ -25,6 +22,11 @@ body {
   background-color:white;
   overflow:hidden;
   display:block;
+  transform-origin:0% 0%;
+  -ms-transform-origin:0% 0%;
+  -moz-transform-origin:0% 0%;
+  -webkit-transform-origin:0% 0%;
+  -o-transform-origin:0% 0%;
 }
 .l {
   position:absolute;

share/hide_pages.js

@@ -1,12 +0,0 @@
-/* hide all pages */
-(function(){
- var s = '.b{display:none;}';
- var n = document.createElement("style");
- n.type = "text/css";
- if (n.styleSheet) {
-  n.styleSheet.cssText = s;
- } else {
-  n.appendChild(document.createTextNode(s));
- }
- document.getElementsByTagName("head")[0].appendChild(n);
-})();

share/manifest

@@ -24,7 +24,7 @@
 @base.css
 $css
 @jquery.js
-@hide_pages.js
+@pdf2htmlEX.js
 
 """
 <title></title>
@@ -37,11 +37,6 @@ $pages
 
 """
 </div>
-"""
-
-@scroll.js
-
-"""
 </body>
 </html>
 """

share/pdf2htmlEX.js

@@ -0,0 +1,91 @@
+/* -*- Mode: Java; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
+/* vim: set shiftwidth=2 tabstop=2 autoindent cindent expandtab: */
+/* vim modeline copied from pdf.js */
+
+/*
+ * pdf2htmlEX.js
+ *
+ * handles UI events/actions/effects
+ *
+ * Copyright 2012 Lu Wang <coolwanglu@gmail.com>
+ */
+
+var pdf2htmlEX = (function(){
+  var obj = {
+    pages : [],
+    page_boxes : [],
+    container : null,
+    selective_render_timer : null,
+
+    init_before_loading_content : function() {
+      /*hide all pages before loading, will reveal only visible ones later */
+      this.hide_pages();
+    },
+
+    init_after_loading_content : function() {
+      this.pages = $(".p");
+      this.page_boxes = $(".b");
+      this.container = $("#pdf-main");
+
+      this.selective_render();
+
+      var _ = this;
+      this.container.scroll(function(){
+        if(_.selective_render_timer)
+          clearTimeout(_.selective_render_timer);
+        _.selective_render_timer = setTimeout(function () {
+          _.selective_render();
+        }, 200);
+      });
+    },
+
+    init : function() {
+      this.init_before_loading_content();
+
+      var _ = this;
+      $(document).ready(function(){_.init_after_loading_content();});
+    },
+
+    hide_pages : function() {
+      /* pages might have not been loaded yet, so add a CSS rule */
+      var s = '.b{display:none;}';
+      var n = document.createElement("style");
+      n.type = "text/css";
+      if (n.styleSheet) {
+        n.styleSheet.cssText = s;
+      } else {
+        n.appendChild(document.createTextNode(s));
+      }
+      document.getElementsByTagName("head")[0].appendChild(n);
+    },
+
+    selective_render : function () {
+      /* hide (positional) invisible pages */
+      var l = this.pages.length;
+      var ch = this.container.height();
+      
+      var i;
+      for(i = 0; i < l; ++i) {
+        var cur_page = $(this.pages[i]);
+
+        if(cur_page.offset().top + cur_page.height() >= 0) break;
+        if(i > 0) $(this.page_boxes[i-1]).hide();
+      }
+
+      if((i > 0) && (i < l)) $(this.page_boxes[i-1]).show();
+
+      for(; i < l; ++i) {
+        $(this.page_boxes[i]).show();
+
+        if($(this.pages[i]).offset().top > ch) break;
+      }
+
+      for(++i; i < l; ++i) {
+        $(this.page_boxes[i]).hide();
+      }
+    },
+  };
+
+  obj.init();
+  return obj;
+})();

share/scroll.js

@@ -1,64 +0,0 @@
-/*
- * scroll.js
- * render only necessary pages
- *
- * by Hongliang TIAN
- * modifiedy by Lu WANG
- */
-$(function() {
-  var $pages = $(".p");
-  var $page_boxes= $(".b");
-  var $main  = $("#pdf-main");
-  var l = $pages.length;
-
-  function isPageVisible(i, H) {
-    var $p = $($pages[i]);
-    var t = $p.offset().top;
-    var b = t + $p.height();
-    return ( ! ( b < 0 || t >= H ) );
-  }
-
-  function setVisibilities(from, to, visible) {
-    for(var i = from; i <= to; ++i) {
-      var $pb = $($page_boxes[i]);
-      if(visible) $pb.show(); else $pb.hide();
-    }
-  }
-
-  // Selectively rendering of pages that are visible or will be visible
-  function selectiveRender() {
-    var first = 0, last = l - 1, H = $main.height();
-
-    // Find the first visible page
-    while(first < l && !isPageVisible(first, H))
-      first++; 
-    // Find the last visible page
-    while(last >= 0 && !isPageVisible(last, H))
-      last--;
-    // Set invisible
-    setVisibilities(first > 0 ? first-1 : first, 
-                    last < l -1 ? last + 1 : last, true);
-    setVisibilities(0, first - 2, false);
-    setVisibilities(last + 2, l-1, false);
-  }
-
-  // Listen to scrolling events to render proper pages
-  var scrolled = false;
-  var last_scroll_time = Date.now();
-  $("#pdf-main").scroll(function() {
-    scrolled = true;
-    last_scroll_time = Date.now();
-  });
-
-  setInterval(function() {
-    // If scrolling pauses 200+ms
-    if (scrolled && (Date.now() - last_scroll_time> 100)) {
-      scrolled = false;
-      // Only render pages that are or will be visible
-      selectiveRender();
-    }
-  }, 100);
-
-  // Trigger the event
-  $("#pdf-main").scroll();
-});

src/HTMLRenderer/export.cc

@@ -97,20 +97,19 @@ void HTMLRenderer::export_transform_matrix (long long tm_id, const double * tm)
     // we have already shifted the origin
     
     // TODO: recognize common matices
-    if(_tm_euqal(tm, id_matrix, 4))
+    if(_tm_equal(tm, id_matrix, 4))
     {
         auto prefixes = {"", "-ms-", "-moz-", "-webkit-", "-o-"};
         for(auto iter = prefixes.begin(); iter != prefixes.end(); ++iter)
-            css_fout << prefix << "transform:none;";
+            css_fout << *iter << "transform:none;";
     }
     else
     {
         auto prefixes = {"", "-ms-", "-moz-", "-webkit-", "-o-"};
         for(auto iter = prefixes.begin(); iter != prefixes.end(); ++iter)
         {
-            const auto & prefix = *iter;
             // PDF use a different coordinate system from Web
-            css_fout << prefix << "transform:matrix("
+            css_fout << *iter << "transform:matrix("
                 << _round(tm[0]) << ','
                 << _round(-tm[1]) << ','
                 << _round(-tm[2]) << ','

src/HTMLRenderer/general.cc

@@ -9,6 +9,7 @@
 
 #include <cstdio>
 #include <ostream>
+#include <cmath>
 
 #include <splash/SplashBitmap.h>
 #include <Link.h>
@@ -24,6 +25,7 @@ namespace pdf2htmlEX {
 using std::fixed;
 using std::flush;
 using std::ostream;
+using std::max;
 
 static void dummy(void *, enum ErrorCategory, int pos, char *)
 {
@@ -47,6 +49,12 @@ HTMLRenderer::HTMLRenderer(const Param * param)
     cur_mapping = new int32_t [0x10000];
     cur_mapping2 = new char* [0x100];
     width_list = new int [0x10000];
+
+    /*
+     * determine scale factors
+     */
+    scale_factor1 = max(param->zoom, param->font_size_multiplier);  
+    scale_factor2 = (param->zoom) / scale_factor1;
 }
 
 HTMLRenderer::~HTMLRenderer()
@@ -113,7 +121,7 @@ void HTMLRenderer::process(PDFDoc *doc)
             }
         }
 
-        doc->displayPage(this, i, param->zoom * DEFAULT_DPI, param->zoom * DEFAULT_DPI,
+        doc->displayPage(this, i, scale_factor1 * DEFAULT_DPI, scale_factor1 * DEFAULT_DPI,
                 0, true, false, false,
                 nullptr, nullptr, nullptr, nullptr);
 
@@ -147,8 +155,18 @@ void HTMLRenderer::startPage(int pageNum, GfxState *state)
     assert((!line_opened) && "Open line in startPage detected!");
 
     html_fout 
-        << "<div id=\"p" << pageNum << "\" class=\"p\" style=\"width:" << pageWidth << "px;height:" << pageHeight << "px;\">"
+        << "<div id=\"p" << pageNum << "\" class=\"p\" style=\"width:" 
-        << "<div id=\"b" << pageNum << "\" class=\"b\" style=\"width:" << pageWidth << "px;height:" << pageHeight << "px;";
+            << (pageWidth * scale_factor2) << "px;height:" 
+            << (pageHeight * scale_factor2) << "px;\">"
+        << "<div id=\"b" << pageNum << "\" class=\"b\" style=\"width:" 
+            << pageWidth << "px;height:" 
+            << pageHeight << "px;";
+
+    {
+        auto prefixes = {"", "-ms-", "-moz-", "-webkit-", "-o-"};
+        for(auto iter = prefixes.begin(); iter != prefixes.end(); ++iter)
+            html_fout << *iter << "transform:scale(" << scale_factor2 << ");";
+    }
 
     if(param->process_nontext)
     {

src/HTMLRenderer/state.cc

@@ -146,7 +146,7 @@ void HTMLRenderer::check_state_change(GfxState * state)
         double new_draw_ctm[6];
         memcpy(new_draw_ctm, cur_ctm, sizeof(new_draw_ctm));
 
-        double new_draw_scale = (param->font_size_multiplier) / (param->zoom) * sqrt(new_draw_ctm[2] * new_draw_ctm[2] + new_draw_ctm[3] * new_draw_ctm[3]);
+        double new_draw_scale = sqrt(new_draw_ctm[2] * new_draw_ctm[2] + new_draw_ctm[3] * new_draw_ctm[3]);
 
         double new_draw_font_size = cur_font_size;
         if(_is_positive(new_draw_scale))

src/include/HTMLRenderer.h

@@ -204,6 +204,18 @@ class HTMLRenderer : public OutputDev
         double pageWidth ;
         double pageHeight ;
 
+        /*
+         * The content of each page is first scaled with factor1 (>=1), then scale back with factor2(<=1)
+         *
+         * factor1 is use to multiplied with all metrics (height/width/font-size...), in order to improve accuracy
+         * factor2 is applied with css transform, and is exposed to Javascript
+         *
+         * factor1 & factor 2 are determined according to zoom and font-size-multiplier
+         *
+         */
+        double scale_factor1;
+        double scale_factor2;
+
 
         ////////////////////////////////////////////////////
         // states