src/html/htmlfilt.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        htmlfilt.cpp
   3 // Purpose:     wxHtmlFilter - input filter for translating into HTML format
   4 // Author:      Vaclav Slavik
   5 // RCS-ID:      $Id$
   6 // Copyright:   (c) 1999 Vaclav Slavik
   7 // Licence:     wxWindows licence
   8 /////////////////////////////////////////////////////////////////////////////
   9
  10 #include "wx/wxprec.h"
  11
  12 #include "wx/defs.h"
  13 #if wxUSE_HTML && wxUSE_STREAMS
  14
  15 #ifdef __BORLANDC__
  16 #pragma hdrstop
  17 #endif
  18
  19 #ifndef WXPRECOMP
  20     #include "wx/log.h"
  21     #include "wx/intl.h"
  22 #endif
  23
  24 #include "wx/strconv.h"
  25 #include "wx/html/htmlfilt.h"
  26 #include "wx/html/htmlwin.h"
  27
  28 // utility function: read a wxString from a wxInputStream
  29 static void ReadString(wxString& str, wxInputStream* s, wxMBConv& conv)
  30 {
  31     size_t streamSize = s->GetSize();
  32
  33     if (streamSize == ~(size_t)0)
  34     {
  35         const size_t bufSize = 4095;
  36         char buffer[bufSize+1];
  37         size_t lastRead;
  38
  39         do
  40         {
  41             s->Read(buffer, bufSize);
  42             lastRead = s->LastRead();
  43             buffer[lastRead] = 0;
  44             str.Append(wxString(buffer, conv));
  45         }
  46         while (lastRead == bufSize);
  47     }
  48     else
  49     {
  50         char* src = new char[streamSize+1];
  51         s->Read(src, streamSize);
  52         src[streamSize] = 0;
  53         str = wxString(src, conv);
  54         delete[] src;
  55     }
  56 }
  57
  58 /*
  59
  60 There is code for several default filters:
  61
  62 */
  63
  64 IMPLEMENT_ABSTRACT_CLASS(wxHtmlFilter, wxObject)
  65
  66 //--------------------------------------------------------------------------------
  67 // wxHtmlFilterPlainText
  68 //          filter for text/plain or uknown
  69 //--------------------------------------------------------------------------------
  70
  71 IMPLEMENT_DYNAMIC_CLASS(wxHtmlFilterPlainText, wxHtmlFilter)
  72
  73 bool wxHtmlFilterPlainText::CanRead(const wxFSFile& WXUNUSED(file)) const
  74 {
  75     return true;
  76 }
  77
  78
  79
  80 wxString wxHtmlFilterPlainText::ReadFile(const wxFSFile& file) const
  81 {
  82     wxInputStream *s = file.GetStream();
  83     wxString doc, doc2;
  84
  85     if (s == NULL) return wxEmptyString;
  86     ReadString(doc, s, wxConvISO8859_1);
  87
  88     doc.Replace(wxT("&"), wxT("&amp;"), true);
  89     doc.Replace(wxT("<"), wxT("&lt;"), true);
  90     doc.Replace(wxT(">"), wxT("&gt;"), true);
  91     doc2 = wxT("<HTML><BODY><PRE>\n") + doc + wxT("\n</PRE></BODY></HTML>");
  92     return doc2;
  93 }
  94
  95
  96
  97
  98
  99 //--------------------------------------------------------------------------------
 100 // wxHtmlFilterImage
 101 //          filter for image/*
 102 //--------------------------------------------------------------------------------
 103
 104 class wxHtmlFilterImage : public wxHtmlFilter
 105 {
 106     DECLARE_DYNAMIC_CLASS(wxHtmlFilterImage)
 107
 108     public:
 109         virtual bool CanRead(const wxFSFile& file) const;
 110         virtual wxString ReadFile(const wxFSFile& file) const;
 111 };
 112
 113 IMPLEMENT_DYNAMIC_CLASS(wxHtmlFilterImage, wxHtmlFilter)
 114
 115
 116
 117 bool wxHtmlFilterImage::CanRead(const wxFSFile& file) const
 118 {
 119     return (file.GetMimeType().Left(6) == wxT("image/"));
 120 }
 121
 122
 123
 124 wxString wxHtmlFilterImage::ReadFile(const wxFSFile& file) const
 125 {
 126     wxString res = wxT("<HTML><BODY><IMG SRC=\"") + file.GetLocation() + wxT("\"></BODY></HTML>");
 127     return res;
 128 }
 129
 130
 131
 132
 133 //--------------------------------------------------------------------------------
 134 // wxHtmlFilterHTML
 135 //          filter for text/html
 136 //--------------------------------------------------------------------------------
 137
 138
 139 IMPLEMENT_DYNAMIC_CLASS(wxHtmlFilterHTML, wxHtmlFilter)
 140
 141 bool wxHtmlFilterHTML::CanRead(const wxFSFile& file) const
 142 {
 143 //    return (file.GetMimeType() == "text/html");
 144 // This is true in most case but some page can return:
 145 // "text/html; char-encoding=...."
 146 // So we use Find instead
 147   return (file.GetMimeType().Find(wxT("text/html")) == 0);
 148 }
 149
 150
 151
 152 wxString wxHtmlFilterHTML::ReadFile(const wxFSFile& file) const
 153 {
 154     wxInputStream *s = file.GetStream();
 155     wxString doc;
 156
 157     if (s == NULL)
 158     {
 159         wxLogError(_("Cannot open HTML document: %s"), file.GetLocation().c_str());
 160         return wxEmptyString;
 161     }
 162
 163     // NB: We convert input file to wchar_t here in Unicode mode, based on
 164     //     either Content-Type header or <meta> tags. In ANSI mode, we don't
 165     //     do it as it is done by wxHtmlParser (for this reason, we add <meta>
 166     //     tag if we used Content-Type header).
 167 #if wxUSE_UNICODE
 168     int charsetPos;
 169     if ((charsetPos = file.GetMimeType().Find(_T("; charset="))) != wxNOT_FOUND)
 170     {
 171         wxString charset = file.GetMimeType().Mid(charsetPos + 10);
 172         wxCSConv conv(charset);
 173         ReadString(doc, s, conv);
 174     }
 175     else
 176     {
 177         wxString tmpdoc;
 178         ReadString(tmpdoc, s, wxConvISO8859_1);
 179         wxString charset = wxHtmlParser::ExtractCharsetInformation(tmpdoc);
 180         if (charset.empty())
 181             doc = tmpdoc;
 182         else
 183         {
 184             wxCSConv conv(charset);
 185             doc = wxString(tmpdoc.mb_str(wxConvISO8859_1), conv);
 186         }
 187     }
 188 #else // !wxUSE_UNICODE
 189     ReadString(doc, s, wxConvLibc);
 190     // add meta tag if we obtained this through http:
 191     if (!file.GetMimeType().empty())
 192     {
 193         wxString hdr;
 194         wxString mime = file.GetMimeType();
 195         hdr.Printf(_T("<meta http-equiv=\"Content-Type\" content=\"%s\">"), mime.c_str());
 196         return hdr+doc;
 197     }
 198 #endif
 199
 200     return doc;
 201 }
 202
 203
 204
 205
 206 ///// Module:
 207
 208 class wxHtmlFilterModule : public wxModule
 209 {
 210     DECLARE_DYNAMIC_CLASS(wxHtmlFilterModule)
 211
 212     public:
 213         virtual bool OnInit()
 214         {
 215             wxHtmlWindow::AddFilter(new wxHtmlFilterHTML);
 216             wxHtmlWindow::AddFilter(new wxHtmlFilterImage);
 217             return true;
 218         }
 219         virtual void OnExit() {}
 220 };
 221
 222 IMPLEMENT_DYNAMIC_CLASS(wxHtmlFilterModule, wxModule)
 223
 224 #endif