1 files changed, 209 insertions, 43 deletions

diff --git a/noncore/apps/opie-reader/striphtml.cpp b/noncore/apps/opie-reader/striphtml.cpp
index e86402b..c434dbb 100644
--- a/noncore/apps/opie-reader/striphtml.cpp
+++ b/noncore/apps/opie-reader/striphtml.cpp
@@ -3,17 +3,16 @@
 #include <qtextstream.h>
 #include <qdir.h>
 #ifdef USEQPE
 #include <qpe/global.h>
 #endif
 #include "CDrawBuffer.h"
 #include "striphtml.h"
 #include "hrule.h"
-#include "util.h"
 
 #include <qregexp.h>
 #include <qimage.h>
 #include <qpixmap.h>
 //#include <qprogressdialog.h>
 //#include <qapplication.h>
 
 static unsigned char h2i(unsigned char c)
@@ -85,25 +84,33 @@ void striphtml::skipblock(const QString& _ent)
       
       ch = skip_ws();
       
       ent = getname(ch, " >").lower();
       qDebug("Skipblock:%s", (const char*)ent);
     } while (ent != _ent && ch != UEOF);
 }
 
-void striphtml::locate(unsigned int n)
+void striphtml::reset()
 {
   m_inblock = false;
   text_q = "";
+  q = "";
+  tablenesteddepth = 0;
   forcecentre = false;
   ignorespace = false;
+  indent = 0;
   while (!stylestack.isEmpty()) stylestack.pop();
   currentstyle.unset();
+}
+
+void striphtml::locate(unsigned int n)
+{
   qDebug("striphtml:locating:%u", n);
+  reset();
   parent->locate(n);
 }
 
 int striphtml::getpara(CBuffer& buff, unsigned long& startpos)
 {
   tchar ch;
   CStyle sty;
   unsigned long pos;
@@ -228,16 +235,23 @@ QString striphtml::dehtml(const QString& _info)
   }
   return info;
 }
 
 bool striphtml::findanchor(const QString& _info)
 {
   //  QProgressDialog dlg("Finding link...", QString::null, 0, NULL, "progress", true);
   //  QProgressBar dlg(0);
+  if (parent->findanchor(_info))
+    {
+      reset();
+      return true;
+    }
+  qDebug("Using html find");
+  parent->locate(parent->startSection());
 #if defined(USEQPE) || defined(_WINDOWS) 
   QString info;
   for (int i = 0; i < _info.length(); i++)
   {
     tchar ch = _info[i];
     if (QString(".^$[]*+?").find(ch) != -1)
       {
         info += '\\';
@@ -290,33 +304,43 @@ bool striphtml::findanchor(const QString& _info)
     {
       locate(startpos);
       qDebug("Not found");
       ret = false;
     }
   return ret;
 }
 
-striphtml::striphtml(const QString& _s) : entmap(NULL), isPre(false), currentid(0), lastch(0), currentfile(_s), indent(0), forcecentre(false), m_inblock(false), m_bchm(false), ignorespace(false)
+striphtml::striphtml(const QString& _s) : entmap(NULL), isPre(false), currentid(0), lastch(0), currentfile(_s), indent(0), forcecentre(false), m_inblock(false), m_bchm(false), ignorespace(false), tablenesteddepth(0)
 {
   href2filepos = new QMap<QString, unsigned long>;
   id2href = new QMap<unsigned long, QString>;
 }
 
 striphtml::~striphtml()
 {
   if (entmap != NULL) delete entmap;
   delete href2filepos;
   delete id2href;
 }
 
 void striphtml::initentmap()
 {
   entmap = new QMap<QString, tchar>;
-  QString fname(QTReaderUtil::getPluginPath("data"));
+#ifdef USEQPE
+#ifdef OPIE
+  QString fname(getenv("OPIEDIR"));
+#else
+  QString fname(getenv("QTDIR"));
+#endif
+  fname += "/plugins/reader/data";
+#else
+  QString fname(getenv("READERDIR"));
+  fname += "/data";
+#endif  
   QFileInfo fi;
   fi.setFile(fname, "HTMLentities");
   if (fi.exists())
     {
       fname = fi.absFilePath();
 
       QFile fl(fname);
       if (fl.open(IO_ReadOnly))
@@ -407,16 +431,22 @@ QString striphtml::getattr(tchar& ch)
     {
       ch = skip_ws();
       if (ch == '"')
         {
           mygetch(ch, sty, pos);
           ref = getname(ch, "\"");
           ch = skip_ws();
         }
+      else if (ch == '\'')
+        {
+          mygetch(ch, sty, pos);
+          ref = getname(ch, "\'");
+          ch = skip_ws();
+        }
       else
         {
           ref = getname(ch, " >");
           if (ch == ' ') ch = skip_ws();
         }
     }
   return ref;
 }
@@ -473,37 +503,41 @@ linkType striphtml::hyperlink(unsigned int n, unsigned int, QString& w, QString&
         }
       
       qDebug("File:%s", (const char*)file);
       qDebug("Name:%s", (const char*)name);
       
 
       if (file.isEmpty())
         {
+          if (parent->findanchor(name))
+            {
+              reset();
+              return eLink;
+            }
           fpit = href2filepos->find(name);
           if (fpit != href2filepos->end())
             {
               locate(*fpit);
               return eLink;
             }
           else
             {
                       //      nm = QString("<a[^>]*name[ \t]*=[ \t]*\"") + name + "\"";
               qDebug("Do a search for:%s", (const char*)name);
-              parent->locate(0);
               findanchor(name);
               return eLink;
             }
         }
       else
         //      if (href.find('#') == -1)
         {
           if (m_bchm)
             {
-              w = "/"+file;
+              w = file;
               nm = name;
               return eFile;
             }
           else
             {
               QFileInfo f(currentfile);
               QFileInfo f1(f.dir(true), file);
               if (f1.exists())
@@ -559,17 +593,41 @@ void striphtml::mygetch(tchar& ch, CStyle& sty, unsigned long& pos)
         }
       if (ch == '>')
         {
           m_inblock = false;
         }
     }
   if (ch == 10 && !isPre)
     {
+#ifdef REMOVE_LF_BEFORE_ENDTAG
+      parent->getch(ch, sty, pos);
+      if (ch == '<')
+        {
+          parent->getch(ch, sty, pos);
+          if (ch == '/')
+            {
+              ch = '<';
+              text_q += '/';
+            }
+          else
+            {
+              text_q += '<';
+              text_q += ch;
+              ch = ' ';
+            }
+        }
+      else
+        {
+          text_q += ch;
+          ch = ' ';
+        }
+#else
       ch = ' ';
+#endif
     }
 }
 
 void striphtml::parse_paragraph(CStyle& currentstyle, tchar& ch, unsigned long pos)
 {
 /*
  int count = 0;
    for (CList<CStyle>::iterator iter = stylestack.begin(); iter != stylestack.end(); ++iter)
@@ -579,17 +637,16 @@ void striphtml::parse_paragraph(CStyle& currentstyle, tchar& ch, unsigned long p
    qDebug("Currently have %u styles", count);
 */
    if (stylestack.isEmpty())
      {
         currentstyle.unset();
      }
    else
      {
-        qDebug("Using stack style");
         currentstyle = stylestack.first();
      }
   if (forcecentre)
     {
       currentstyle.setCentreJustify();
     }
   if (ch == ' ') ch = skip_ws();
   while (ch != '>' && ch != UEOF)
@@ -602,16 +659,20 @@ void striphtml::parse_paragraph(CStyle& currentstyle, tchar& ch, unsigned long p
           if (attr == "center")
             {
               currentstyle.setCentreJustify();
             }
           if (attr == "right")
             {
               currentstyle.setRightJustify();
             }
+          if (attr == "justify")
+            {
+              currentstyle.setFullJustify();
+            }
         }
       if (ent == "id")
         {
           (*href2filepos)[attr] = pos;
         }
       if (ent == "bgcolor")
         {
           qDebug("Got paper colour:%s", (const char*)attr);
@@ -649,44 +710,43 @@ void striphtml::getch(tchar& ch, CStyle& sty, unsigned long& pos)
           currentstyle.setPicture(false, hRule(w,h,r,g,b));
           q = q.right(q.length()-6);
         }
       else
         {
           q = q.right(q.length()-1);
         }
       sty = currentstyle;
+      lastch = ch;
       return;
     }
   do
     {
       unsigned long npos;
       CStyle dummy;
       mygetch(ch, dummy, pos);
-      //    if (ch == 10 && !isPre) ch = ' ';
       while (ch == '<' && ch != UEOF)
         {
           ch = skip_ws();
-
           QString ent = getname(ch, " >").lower();
 
                   //  qDebug("Entity:%s", (const char*)ent);
         
           if (ent == "a"/* || ent == "reference"*/)
             {
               if (ch == ' ') ch = skip_ws();
               bool fileposfound = false;
               bool ishref = false;
               unsigned int filepos = 0;
               QString ref, name;
               while (ch != '>' && ch != UEOF)
                 {
                   QString ent = getname(ch, " =>").lower();
                   QString attr = getattr(ch);
-                  qDebug("<A>Entity:%s Attr:%s", (const char*)ent, (const char*)attr);
+                  //qDebug("<A>Entity:%s Attr:%s", (const char*)ent, (const char*)attr);
                   if (ent == "name")
                     {
                       name = attr;
                     }
                    if (ent == "onclick")
                      {
                         int st = attr.find('\'');
                         int nd = attr.findRev('\'');
@@ -707,17 +767,17 @@ void striphtml::getch(tchar& ch, CStyle& sty, unsigned long& pos)
                           ishref = true;
                           ref = attr;
                         }
                     }
                   if (ent == "title")
                     {
                       text_q = attr + "</a><p>";
                     }
-                  qDebug("<a %s=%s>", (const char*)ent, (const char*)ref);
+                  //qDebug("<a %s=%s>", (const char*)ent, (const char*)ref);
                 }
               if (ishref)
                 {
                   currentstyle.setColour(0,0,255);
                   currentstyle.setLink(true);
                   currentstyle.setData(currentid);
                   if (!text_q.isEmpty())
                     {
@@ -742,18 +802,19 @@ void striphtml::getch(tchar& ch, CStyle& sty, unsigned long& pos)
             {
               parse_paragraph(currentstyle, ch, pos);
               currentstyle.setExtraSpace(3);
               continue;
             }
           else if (ent == "div")
             {
               parse_paragraph(currentstyle, ch, pos);
-               stylestack.push_front(currentstyle);
-              //indent = 0;
+              stylestack.push_front(currentstyle);
+              currentstyle.setExtraSpace(16);
+                  //indent = 0;
               continue;
             }
           else if (ent == "sup")
             {
               currentstyle.setVOffset(-1);
             }
           else if (ent == "sup")
             {
@@ -795,16 +856,17 @@ void striphtml::getch(tchar& ch, CStyle& sty, unsigned long& pos)
                if (!stylestack.isEmpty())
                  {
                     stylestack.pop();
                  }
             }
           else if (ent == "pre")
             {
               isPre = true;
+              currentstyle.setNoJustify();
               currentstyle.setMono();
             }
           else if (ent == "tt")
             {
               currentstyle.setMono();
             }
           else if (ent == "b" || ent == "strong")
             {
@@ -817,24 +879,27 @@ void striphtml::getch(tchar& ch, CStyle& sty, unsigned long& pos)
           else if (ent == "/u")
             {
               currentstyle.unsetUnderline();
             }
           else if (ent == "blockquote")
             {
               if (ch != '>') ch = skip_ws_end();
               ch = 10;
+              currentstyle.setExtraSpace(0);
               currentstyle.setLeftMargin(30);
               currentstyle.setRightMargin(30);
               continue;
             }
           else if (ent == "br" || ent == "br/")
             {
               if (ch != '>') ch = skip_ws_end();
               ch = 10;
+              currentstyle.setExtraSpace(0);
+              lastch = 0;
               continue;
             }
           else if (ent == "mbp:pagebreak")
             {
               /*
               if (ch != '>') ch = skip_ws_end(pos);
               q += 10;
               q += QChar(UEOF);
@@ -856,16 +921,17 @@ void striphtml::getch(tchar& ch, CStyle& sty, unsigned long& pos)
           else if (ent == "/center")
             {
               qDebug("unsetting centre");
               forcecentre = false;
             }
           else if (ent == "li")
             {
               if (ch != '>') ch = skip_ws_end();
+              lastch = 0;
               ch = 10;
               if (m_listtype[indent % m_cmaxdepth] == 1)
                 {
                   q.setNum(m_ctr[indent % m_cmaxdepth]++);
                 }
               else
                 {
                   q += QChar(8226);
@@ -913,54 +979,58 @@ void striphtml::getch(tchar& ch, CStyle& sty, unsigned long& pos)
           else if (ent == "big")
             {
               currentstyle.setFontSize(2);
             }
           else if (ent == "/big")
             {
               currentstyle.setFontSize(0);
             }
+          else if (ent[0] == '/' && ent[1] == 'h' && ent.length() == 3 && QString("123456789").find(ent[2]) != -1)
+            {
+              parse_paragraph(currentstyle, ch, pos);
+              currentstyle.setExtraSpace(3);
+              continue;
+            }
           else if (ent[0] == 'h' && ent.length() == 2 && QString("123456789").find(ent[1]) != -1)
             {
               indent = 0;
               if (ent[1] == '1')
                 {
                   parse_paragraph(currentstyle, ch, pos);
                   currentstyle.setFontSize(3);
                   currentstyle.setExtraSpace(8);
                   currentstyle.setBold();
                           //        currentstyle.setExtraSpace(10);
-                  continue;
                 }
               else if (ent[1] == '2')
                 {
                   parse_paragraph(currentstyle, ch, pos);
                   currentstyle.setFontSize(2);
                   currentstyle.setExtraSpace(6);
                   currentstyle.setBold();
                           //        currentstyle.setExtraSpace(10);
-                  continue;
                 }
               else if (ent[1] == '3')
                 {
                   parse_paragraph(currentstyle, ch, pos);
                   currentstyle.setFontSize(1);
                   currentstyle.setExtraSpace(4);
                   currentstyle.setBold();
                           //        currentstyle.setExtraSpace(10);
-                  continue;
                 }
               else
                 {
                   parse_paragraph(currentstyle, ch, pos);
                   currentstyle.setExtraSpace(4);
                   currentstyle.setBold();
                           //        currentstyle.setExtraSpace(10);
-                  continue;
                 }
+              ch = 10;
+              continue;
             }
         
         
           else if (ent == "/a")
             {
               currentstyle.setColour(0,0,0);
               currentstyle.setLink(false);
             }
@@ -1042,100 +1112,142 @@ void striphtml::getch(tchar& ch, CStyle& sty, unsigned long& pos)
                        //       ch = 10;
                continue;
             }
           */
           else if (ent[0] == '/' && ent.length() == 3 && ent[1] == 'h' && QString("123456789").find(ent[2]) != -1)
             {
               currentstyle.unset();
               if (ch != '>') ch = skip_ws_end();
-              ch = 10;
-              continue;
+              //ch = 10;
+              //continue;
             }
-
-
-
           else if (ent == "table" || ent == "/table")
             {
               currentstyle.unset();
               ignorespace = (ent == "table");
+              if (ent == "table")
+                {
+                  if (tablenesteddepth++ == 0) currentstyle.setTable(pos);
+                }
+              else
+                {
+                  if (--tablenesteddepth <= 0)
+                    {
+                      tablenesteddepth = 0;
+                      currentstyle.setTable(0xffffffff);
+                    }
+                }
               if (ch == ' ') ch = skip_ws();
               while (ch != '>' && ch != UEOF)
                 {
                   QString ent = getname(ch, " =>").lower();
                   QString attr = getattr(ch);
                   qDebug("<table>Entity:%s Attr:%s", (const char*)ent, (const char*)attr);
                 }
               if (ch != '>') ch = skip_ws_end();
 
+              currentstyle.setLeftMargin(6*tablenesteddepth);
+
+
+              lastch = 0; // Anything but 10
               ch = 10;
               q += '-';
               q += QChar(parent->getwidth());
-              q += 2;
+              q += 3;
               q += '\0';
               q += '\0';
               q += '\0';
               continue;
             }
           else if (ent == "hr")
             {
+              //bool isPageBreak = false;
               if (ch == ' ') ch = skip_ws();
-                      //      if (stylestack.isEmpty())
-                      //        {
-                  currentstyle.unset();
-                          //    }
-                  /*
-              else
-                {
-                  qDebug("Using stack style");
-                  currentstyle = stylestack.first();
-                }
-                  */
               unsigned char red = 0, green = 0, blue = 0;
               while (ch != '>' && ch != UEOF)
                 {
                   QString ent = getname(ch, " =>").lower();
                   QString attr = getattr(ch);
                   if (ent == "color")
                     {
                       parse_color(attr, red, green, blue);
                     }
+                  /*
+                  if (ent == "size")
+                    {
+                      if (attr == "0")
+                        {
+                          isPageBreak = true;
+                        }
+                    }
+                  */
                   qDebug("<hr>Entity:%s Attr:%s", (const char*)ent, (const char*)attr);
                 }
               if (ch != '>') ch = skip_ws_end();
-              ch = 10;
-              q += '-';
-              q += QChar(parent->getwidth());
-              q += 3;
-              q += red;
-              q += green;
-              q += blue;
+              /*
+              if (isPageBreak)
+                {
+                  ch = UEOF;
+                }
+              else
+                {
+              */
+                          //      if (stylestack.isEmpty())
+                          //    {
+                  currentstyle.unset();
+                          //    }
+                  /*
+                    else
+                    {
+                    qDebug("Using stack style");
+                    currentstyle = stylestack.first();
+                    }
+                  */
+                  lastch = 0; //Anything but 10 or ' '
+                  ch = 10;
+                  q += '-';
+                  q += QChar(parent->getwidth());
+                  q += 3;
+                  q += red;
+                  q += green;
+                  q += blue;
+
               continue;
             }
 
 
 
           else if (ent == "img")
             {
               if (ch == ' ') ch = skip_ws();
               while (ch != '>' && ch != UEOF)
                 {
                   QString ent = getname(ch, " =>").lower();
                   QString attr = getattr(ch);
                   qDebug("<img>Entity:%s Attr:%s", (const char*)ent, (const char*)attr);
                   if (ent == "src")
                     {
+                      /*
                       if (m_bchm)
                         {
                           QImage* img = parent->getPicture(attr);
                           if (img != NULL)
                             {
                               currentstyle.setPicture(true, img);
                             }
                         }
+                      */
+
+
+                      QImage* img = parent->getPicture(attr);
+                      if (img != NULL)
+                        {
+                          currentstyle.setPicture(true, img);
+                        }
                       else
                         {
                           QFileInfo f(currentfile);
                           QFileInfo f1(f.dir(true), attr);
                           QPixmap pm;
                           if (pm.load(f1.absFilePath()))
                             {
                               QImage* img = new QImage(pm.convertToImage());
@@ -1167,16 +1279,24 @@ void striphtml::getch(tchar& ch, CStyle& sty, unsigned long& pos)
             {
               QString nd("/");
               skipblock(nd+ent);
             }
           else if (ent == "metadata")
             {
                       //      skipblock("/metadata");
             }
+          else if (ent == "title")
+            {
+              skipblock("/title");
+            }
+          else if (ent == "head")
+            {
+              skipblock("/head");
+            }
           /*
           else if (ent == "metadata")
             {
               currentstyle.setFontSize(-2);
             }
           else if (ent == "/metadata")
             {
               currentstyle.unset();
@@ -1186,17 +1306,20 @@ void striphtml::getch(tchar& ch, CStyle& sty, unsigned long& pos)
           */
           else
             {
               if (ent[0] != '/')
                 qDebug("Not handling:%s", (const char*)ent);
             }
 
           if (ch != '>') ch = skip_ws_end();
-          mygetch(ch, dummy, npos);
+          if (ent[0] == '/')
+            mygetch(ch, dummy, pos);
+          else
+            mygetch(ch, dummy, npos);
         }
       if (ch == '&')
         {
           mygetch(ch, dummy, npos);
           if (ch == '#')
             {
               int id = 0;
               mygetch(ch, dummy, npos);
@@ -1229,20 +1352,63 @@ void striphtml::getch(tchar& ch, CStyle& sty, unsigned long& pos)
                 }
               else
                 {
                   ch = '.';
                 }
             }
         }
       //    sty = (dummy == ucFontBase) ? currentstyle : dummy;
+      if (lastch == 10 && ch == 10 && sty.getExtraSpace() > currentstyle.getExtraSpace())
+        {
+          currentstyle.setExtraSpace(sty.getExtraSpace());
+        }
       sty = currentstyle;
     }
-  while (!isPre && (lastch == ' '  || lastch == 10 || ignorespace) && ch == ' ');
+  while (!isPre && (((lastch == ' '  || lastch == 10 || ignorespace) && ch == ' ') || ((ch == 10) && (lastch == 10))));
   //  lastch = ch;
   lastch = ch;
   return;
 }
 
+QString striphtml::getTableAsHtml(unsigned long loc)
+{
+  qDebug("striphtml::getTableAsHtml");
+  QString ret;
+  tchar ch(0);
+  CStyle sty;
+  unsigned long pos;
+  locate(loc);
+  int endpos(0);
+  QString endmarker("</table>");
+  QString startmarker("<table");
+  int startpos(0);
+  int depth(0);
+  while (ch != UEOF)
+    {
+      parent->getch(ch, sty, pos);
+      QChar qc(ch);
+      ret += qc;
+      if (qc.lower() == endmarker[endpos])
+        {
+          if ((++endpos >= endmarker.length()) && (--depth <= 0)) break;
+        }
+      else
+        {
+          endpos = 0;
+        }
+      if (qc.lower() == startmarker[startpos])
+        {
+          if (++startpos >= startmarker.length()) ++depth;
+        }
+      else
+        {
+          startpos = 0;
+        }
+    }
+  return ret;
+}
+
+
 extern "C"
 {
   CFilter* newfilter(const QString& s) { return new striphtml(s); }
 }