process outline

CMakeLists.txt

@@ -154,6 +154,7 @@ add_executable(pdf2htmlEX
     src/HTMLRenderer/TextLineBuffer.h
     src/HTMLRenderer/TextLineBuffer.cc
     src/HTMLRenderer/link.cc
+    src/HTMLRenderer/outline.cc
     src/HTMLRenderer/state.cc
     src/HTMLRenderer/text.cc
     src/BackgroundRenderer/BackgroundRenderer.h 

share/base.css

@@ -1,10 +1,18 @@
 /* Base CSS */
 /* Copyright 2012 Lu Wang <coolwanglu@gmail.com> */
-#pdf-main { /* PDF container */
+#pdf-outline { /* PDF Outline */
   position:absolute;
   top:0;
   left:0;
   bottom:0;
+  width:200px;
+  overflow:auto;
+}
+#pdf-main { /* PDF container */
+  position:absolute;
+  top:0;
+  left:200px;
+  bottom:0;
   right:0;
   overflow:auto;
   background-color:grey;

share/pdf2htmlEX.js

@@ -129,7 +129,8 @@ var pdf2htmlEX = (function(){
 
       //this.zoom_fixer();
       
-      this.container.on('click', '.a', this, this.annot_link_handler);
+      // used by outline/annot_link etc
+      this.container.on('click', '.a', this, this.link_handler);
 
       this.render();
     },
@@ -231,7 +232,7 @@ var pdf2htmlEX = (function(){
       return this.pages[(new Page(obj.closest('.p')[0])).n];
     },
 
-    annot_link_handler : function (e) {
+    link_handler : function (e) {
       var _ = e.data;
       var t = $(e.currentTarget);
       var cur_page = _.get_containing_page(t);

src/HTMLRenderer/HTMLRenderer.h

@@ -212,13 +212,16 @@ class HTMLRenderer : public OutputDev
         void pre_process(PDFDoc * doc);
         void post_process();
 
+        void process_outline();
+        void process_outline_items(GooList * items);
+
         void set_stream_flags (std::ostream & out);
 
         std::string dump_embedded_font (GfxFont * font, long long fn_id);
         void embed_font(const std::string & filepath, GfxFont * font, FontInfo & info, bool get_metric_only = false);
 
-        // convert a LinkDest to a string that our Javascript code can understand
-        std::string get_linkdest_str(int & pageno, LinkDest * dest);
+        // convert a LinkAction to a string that our Javascript code can understand
+        std::string get_linkaction_str(LinkAction *, std::string & detail);
 
         ////////////////////////////////////////////////////
         // manage styles

src/HTMLRenderer/general.cc

@@ -74,6 +74,9 @@ void HTMLRenderer::process(PDFDoc *doc)
 
     pre_process(doc);
 
+    ///////////////////
+    // Process pages
+    
     BackgroundRenderer * bg_renderer = nullptr;
     if(param->process_nontext)
     {
@@ -120,6 +123,10 @@ void HTMLRenderer::process(PDFDoc *doc)
         cerr << "Working: " << page_count << "/" << page_count;
     cerr << endl;
 
+    ////////////////////////
+    // Process Outline
+    process_outline(); 
+
     post_process();
 
     if(bg_renderer)

src/HTMLRenderer/link.cc

@@ -29,14 +29,16 @@ using std::endl;
 /*
  * The detailed rectangle area of the link destination
  * Will be parsed and performed by Javascript
+ * The string will be put into a HTML attribute, surrounded by single quotes
+ * So pay attention to the characters used here
  */
-string HTMLRenderer::get_linkdest_str(int & pageno, LinkDest * dest)
+static string get_linkdest_detail_str(LinkDest * dest, Catalog * catalog, int & pageno)
 {
     pageno = 0;
     if(dest->isPageRef())
     {
         auto pageref = dest->getPageRef();
-        pageno = cur_catalog->findPage(pageref.num, pageref.gen);
+        pageno = catalog->findPage(pageref.num, pageref.gen);
     }
     else
     {
@@ -125,15 +127,10 @@ string HTMLRenderer::get_linkdest_str(int & pageno, LinkDest * dest)
     return sout.str();
 }
 
-/*
- * Based on pdftohtml from poppler
- * TODO: CSS for link rectangles
- * TODO: share rectangle draw with css-draw
- */
-void HTMLRenderer::processLink(AnnotLink * al)
+string HTMLRenderer::get_linkaction_str(LinkAction * action, string & detail)
 {
-    std::string dest_str, dest_detail_str;
-    auto action = al->getAction();
+    string dest_str;
+    detail = "";
     if(action)
     {
         auto kind = action->getKind();
@@ -150,7 +147,7 @@ void HTMLRenderer::processLink(AnnotLink * al)
                     if(dest)
                     {
                         int pageno = 0;
-                        dest_detail_str = get_linkdest_str(pageno, dest);
+                        detail = get_linkdest_detail_str(dest, cur_catalog, pageno);
                         if(pageno > 0)
                         {
                             dest_str = (char*)str_fmt("#p%x", pageno);
@@ -181,6 +178,19 @@ void HTMLRenderer::processLink(AnnotLink * al)
         }
     }
 
+    return dest_str;
+}
+    
+/*
+ * Based on pdftohtml from poppler
+ * TODO: CSS for link rectangles
+ * TODO: share rectangle draw with css-draw
+ */
+void HTMLRenderer::processLink(AnnotLink * al)
+{
+    string dest_detail_str;
+    string dest_str = get_linkaction_str(al->getAction(), dest_detail_str);
+
     if(!dest_str.empty())
     {
         f_pages.fs << "<a class=\"a\" href=\"" << dest_str << "\"";

src/HTMLRenderer/outline.cc

@@ -0,0 +1,72 @@
+/*
+ * outline.cc
+ *
+ * Handling Outline items
+ *
+ * by WangLu
+ * 2013.01.28
+ */
+
+#include <iostream>
+
+#include <Outline.h>
+#include <goo/GooList.h>
+
+#include "HTMLRenderer.h"
+#include "util/namespace.h"
+#include "util/unicode.h"
+
+namespace pdf2htmlEX {
+
+using std::ostream;
+
+void HTMLRenderer::process_outline_items(GooList * items)
+{
+    if((!items) || (items->getLength() == 0))
+        return;
+
+    f_outline.fs << "<ul>";
+
+    for(int i = 0; i < items->getLength(); ++i)
+    {
+        OutlineItem * item = (OutlineItem*)(items->get(i));
+
+        string detail;
+        string dest = get_linkaction_str(item->getAction(), detail);
+
+        // we don't care dest is empty or not.
+        f_outline.fs << "<li>"
+            << "<a href=\"" << dest << "\"";
+
+        if(!detail.empty())
+            f_outline.fs << " data-dest-detail='" << detail << "'";
+
+        f_outline.fs << ">";
+
+        outputUnicodes(f_outline.fs, item->getTitle(), item->getTitleLength());
+
+        f_outline.fs << "</a>";
+
+        // check kids
+        item->open();
+        if(item->hasKids())
+        {
+            process_outline_items(item->getKids());
+        }
+        item->close();
+        f_outline.fs << "</li>";
+    }
+
+    f_outline.fs << "</ul>";
+}
+   
+void HTMLRenderer::process_outline()
+{
+    Outline * outline = cur_doc->getOutline();
+    if(!outline)
+        return;
+
+    process_outline_items(outline->getItems());
+}
+
+}// namespace pdf2htmlEX

src/util/unicode.h

@@ -33,6 +33,9 @@ Unicode unicode_from_font (CharCode code, GfxFont * font);
  */
 Unicode check_unicode(Unicode * u, int len, CharCode code, GfxFont * font);
 
+/*
+ * Escape necessary characters, and map Unicode to UTF-8
+ */
 void outputUnicodes(std::ostream & out, const Unicode * u, int uLen);