[wxWidgets.git] / src / html / htmlfilt.cpp

/////////////////////////////////////////////////////////////////////////////
// Name:        src/html/htmlfilt.cpp
// Purpose:     wxHtmlFilter - input filter for translating into HTML format
// Author:      Vaclav Slavik
// RCS-ID:      $Id$
// Copyright:   (c) 1999 Vaclav Slavik
// Licence:     wxWindows licence
/////////////////////////////////////////////////////////////////////////////

#include "wx/wxprec.h"

#ifdef __BORLANDC__
    #pragma hdrstop
#endif

#if wxUSE_HTML && wxUSE_STREAMS

#ifndef WXPRECOMP
    #include "wx/log.h"
    #include "wx/intl.h"
#endif

#include "wx/strconv.h"
#include "wx/html/htmlfilt.h"
#include "wx/html/htmlwin.h"

// utility function: read a wxString from a wxInputStream
static void ReadString(wxString& str, wxInputStream* s, wxMBConv& conv)
{
    size_t streamSize = s->GetSize();

    if (streamSize == ~(size_t)0)
    {
        const size_t bufSize = 4095;
        char buffer[bufSize+1];
        size_t lastRead;

        do
        {
            s->Read(buffer, bufSize);
            lastRead = s->LastRead();
            buffer[lastRead] = 0;
            str.Append(wxString(buffer, conv));
        }
        while (lastRead == bufSize);
    }
    else
    {
        char* src = new char[streamSize+1];
        s->Read(src, streamSize);
        src[streamSize] = 0;
        str = wxString(src, conv);
        delete[] src;
    }
}

/*

There is code for several default filters:

*/

IMPLEMENT_ABSTRACT_CLASS(wxHtmlFilter, wxObject)

//--------------------------------------------------------------------------------
// wxHtmlFilterPlainText
//          filter for text/plain or uknown
//--------------------------------------------------------------------------------

IMPLEMENT_DYNAMIC_CLASS(wxHtmlFilterPlainText, wxHtmlFilter)

bool wxHtmlFilterPlainText::CanRead(const wxFSFile& WXUNUSED(file)) const
{
    return true;
}


wxString wxHtmlFilterPlainText::ReadFile(const wxFSFile& file) const
{
    wxInputStream *s = file.GetStream();
    wxString doc, doc2;

    if (s == NULL) return wxEmptyString;
    ReadString(doc, s, wxConvISO8859_1);

    doc.Replace(wxT("&"), wxT("&amp;"), true);
    doc.Replace(wxT("<"), wxT("&lt;"), true);
    doc.Replace(wxT(">"), wxT("&gt;"), true);
    doc2 = wxT("<HTML><BODY><PRE>\n") + doc + wxT("\n</PRE></BODY></HTML>");
    return doc2;
}


//--------------------------------------------------------------------------------
// wxHtmlFilterImage
//          filter for image/*
//--------------------------------------------------------------------------------

class wxHtmlFilterImage : public wxHtmlFilter
{
    DECLARE_DYNAMIC_CLASS(wxHtmlFilterImage)

    public:
        virtual bool CanRead(const wxFSFile& file) const;
        virtual wxString ReadFile(const wxFSFile& file) const;
};

IMPLEMENT_DYNAMIC_CLASS(wxHtmlFilterImage, wxHtmlFilter)


bool wxHtmlFilterImage::CanRead(const wxFSFile& file) const
{
    return (file.GetMimeType().Left(6) == wxT("image/"));
}


wxString wxHtmlFilterImage::ReadFile(const wxFSFile& file) const
{
    wxString res = wxT("<HTML><BODY><IMG SRC=\"") + file.GetLocation() + wxT("\"></BODY></HTML>");
    return res;
}


//--------------------------------------------------------------------------------
// wxHtmlFilterHTML
//          filter for text/html
//--------------------------------------------------------------------------------


IMPLEMENT_DYNAMIC_CLASS(wxHtmlFilterHTML, wxHtmlFilter)

bool wxHtmlFilterHTML::CanRead(const wxFSFile& file) const
{
//    return (file.GetMimeType() == "text/html");
// This is true in most case but some page can return:
// "text/html; char-encoding=...."
// So we use Find instead
  return (file.GetMimeType().Find(wxT("text/html")) == 0);
}


wxString wxHtmlFilterHTML::ReadFile(const wxFSFile& file) const
{
    wxInputStream *s = file.GetStream();
    wxString doc;

    if (s == NULL)
    {
        wxLogError(_("Cannot open HTML document: %s"), file.GetLocation().c_str());
        return wxEmptyString;
    }

    // NB: We convert input file to wchar_t here in Unicode mode, based on
    //     either Content-Type header or <meta> tags. In ANSI mode, we don't
    //     do it as it is done by wxHtmlParser (for this reason, we add <meta>
    //     tag if we used Content-Type header).
#if wxUSE_UNICODE
    int charsetPos;
    if ((charsetPos = file.GetMimeType().Find(_T("; charset="))) != wxNOT_FOUND)
    {
        wxString charset = file.GetMimeType().Mid(charsetPos + 10);
        wxCSConv conv(charset);
        ReadString(doc, s, conv);
    }
    else
    {
        wxString tmpdoc;
        ReadString(tmpdoc, s, wxConvISO8859_1);
        wxString charset = wxHtmlParser::ExtractCharsetInformation(tmpdoc);
        if (charset.empty())
            doc = tmpdoc;
        else
        {
            wxCSConv conv(charset);
            doc = wxString(tmpdoc.mb_str(wxConvISO8859_1), conv);
        }
    }
#else // !wxUSE_UNICODE
    ReadString(doc, s, wxConvLibc);
    // add meta tag if we obtained this through http:
    if (!file.GetMimeType().empty())
    {
        wxString hdr;
        wxString mime = file.GetMimeType();
        hdr.Printf(_T("<meta http-equiv=\"Content-Type\" content=\"%s\">"), mime.c_str());
        return hdr+doc;
    }
#endif

    return doc;
}


///// Module:

class wxHtmlFilterModule : public wxModule
{
    DECLARE_DYNAMIC_CLASS(wxHtmlFilterModule)

    public:
        virtual bool OnInit()
        {
            wxHtmlWindow::AddFilter(new wxHtmlFilterHTML);
            wxHtmlWindow::AddFilter(new wxHtmlFilterImage);
            return true;
        }
        virtual void OnExit() {}
};

IMPLEMENT_DYNAMIC_CLASS(wxHtmlFilterModule, wxModule)

#endif
Commit	Line	Data
5526e819	1	/////////////////////////////////////////////////////////////////////////////
93763ad5	2	// Name: src/html/htmlfilt.cpp
5526e819 VS	3	// Purpose: wxHtmlFilter - input filter for translating into HTML format
5526e819 VS	4	// Author: Vaclav Slavik
69941f05	5	// RCS-ID: $Id$
5526e819	6	// Copyright: (c) 1999 Vaclav Slavik
65571936	7	// Licence: wxWindows licence
5526e819 VS	8	/////////////////////////////////////////////////////////////////////////////
5526e819 VS	9
4dcaf11a	10	#include "wx/wxprec.h"
5526e819	11
2b5f62a0	12	#ifdef __BORLANDC__
93763ad5	13	#pragma hdrstop
5526e819 VS	14	#endif
5526e819 VS	15
93763ad5 WS	16	#if wxUSE_HTML && wxUSE_STREAMS
93763ad5 WS	17
5526e819	18	#ifndef WXPRECOMP
04dbb646 VZ	19	#include "wx/log.h"
04dbb646 VZ	20	#include "wx/intl.h"
5526e819 VS	21	#endif
5526e819 VS	22
2b5f62a0	23	#include "wx/strconv.h"
69941f05	24	#include "wx/html/htmlfilt.h"
4dcaf11a	25	#include "wx/html/htmlwin.h"
5526e819	26
eb37e1d2	27	// utility function: read a wxString from a wxInputStream
2b5f62a0	28	static void ReadString(wxString& str, wxInputStream* s, wxMBConv& conv)
eb37e1d2 MB	29	{
	30	size_t streamSize = s->GetSize();
	31
2b5f62a0	32	if (streamSize == ~(size_t)0)
eb37e1d2 MB	33	{
	34	const size_t bufSize = 4095;
	35	char buffer[bufSize+1];
	36	size_t lastRead;
	37
	38	do
	39	{
	40	s->Read(buffer, bufSize);
	41	lastRead = s->LastRead();
	42	buffer[lastRead] = 0;
2b5f62a0	43	str.Append(wxString(buffer, conv));
eb37e1d2	44	}
2b5f62a0	45	while (lastRead == bufSize);
eb37e1d2 MB	46	}
	47	else
	48	{
	49	char* src = new char[streamSize+1];
	50	s->Read(src, streamSize);
	51	src[streamSize] = 0;
2b5f62a0 VZ	52	str = wxString(src, conv);
2b5f62a0 VZ	53	delete[] src;
eb37e1d2 MB	54	}
eb37e1d2 MB	55	}
5526e819 VS	56
	57	/*
	58
	59	There is code for several default filters:
	60
	61	*/
	62
	63	IMPLEMENT_ABSTRACT_CLASS(wxHtmlFilter, wxObject)
	64
	65	//--------------------------------------------------------------------------------
	66	// wxHtmlFilterPlainText
	67	// filter for text/plain or uknown
	68	//--------------------------------------------------------------------------------
	69
	70	IMPLEMENT_DYNAMIC_CLASS(wxHtmlFilterPlainText, wxHtmlFilter)
	71
a4c97004	72	bool wxHtmlFilterPlainText::CanRead(const wxFSFile& WXUNUSED(file)) const
5526e819	73	{
d1da8872	74	return true;
5526e819 VS	75	}
	76
	77
	78
420ec58a	79	wxString wxHtmlFilterPlainText::ReadFile(const wxFSFile& file) const
5526e819 VS	80	{
5526e819 VS	81	wxInputStream *s = file.GetStream();
5526e819 VS	82	wxString doc, doc2;
	83
	84	if (s == NULL) return wxEmptyString;
2b5f62a0	85	ReadString(doc, s, wxConvISO8859_1);
5526e819	86
d1da8872 WS	87	doc.Replace(wxT("&"), wxT("&"), true);
	88	doc.Replace(wxT("<"), wxT("<"), true);
	89	doc.Replace(wxT(">"), wxT(">"), true);
2b5f62a0	90	doc2 = wxT("<HTML><BODY><PRE>\n") + doc + wxT("\n</PRE></BODY></HTML>");
5526e819 VS	91	return doc2;
	92	}
	93
	94
	95
	96
	97
	98	//--------------------------------------------------------------------------------
	99	// wxHtmlFilterImage
	100	// filter for image/*
	101	//--------------------------------------------------------------------------------
	102
	103	class wxHtmlFilterImage : public wxHtmlFilter
	104	{
	105	DECLARE_DYNAMIC_CLASS(wxHtmlFilterImage)
	106
	107	public:
420ec58a VS	108	virtual bool CanRead(const wxFSFile& file) const;
420ec58a VS	109	virtual wxString ReadFile(const wxFSFile& file) const;
5526e819 VS	110	};
	111
	112	IMPLEMENT_DYNAMIC_CLASS(wxHtmlFilterImage, wxHtmlFilter)
	113
	114
	115
420ec58a	116	bool wxHtmlFilterImage::CanRead(const wxFSFile& file) const
5526e819	117	{
0413cec5	118	return (file.GetMimeType().Left(6) == wxT("image/"));
5526e819 VS	119	}
	120
	121
	122
420ec58a	123	wxString wxHtmlFilterImage::ReadFile(const wxFSFile& file) const
5526e819	124	{
2b5f62a0 VZ	125	wxString res = wxT("<HTML><BODY><IMG SRC=\"") + file.GetLocation() + wxT("\"></BODY></HTML>");
2b5f62a0 VZ	126	return res;
5526e819 VS	127	}
	128
	129
	130
	131
	132	//--------------------------------------------------------------------------------
2b5f62a0 VZ	133	// wxHtmlFilterHTML
2b5f62a0 VZ	134	// filter for text/html
5526e819 VS	135	//--------------------------------------------------------------------------------
5526e819 VS	136
5526e819 VS	137
	138	IMPLEMENT_DYNAMIC_CLASS(wxHtmlFilterHTML, wxHtmlFilter)
	139
420ec58a	140	bool wxHtmlFilterHTML::CanRead(const wxFSFile& file) const
5526e819	141	{
f61815af GL	142	// return (file.GetMimeType() == "text/html");
	143	// This is true in most case but some page can return:
	144	// "text/html; char-encoding=...."
	145	// So we use Find instead
77611ad4	146	return (file.GetMimeType().Find(wxT("text/html")) == 0);
5526e819 VS	147	}
	148
	149
	150
420ec58a	151	wxString wxHtmlFilterHTML::ReadFile(const wxFSFile& file) const
5526e819 VS	152	{
5526e819 VS	153	wxInputStream *s = file.GetStream();
5526e819 VS	154	wxString doc;
5526e819 VS	155
04dbb646	156	if (s == NULL)
f3c82859	157	{
f6bcfd97	158	wxLogError(_("Cannot open HTML document: %s"), file.GetLocation().c_str());
f3c82859 VS	159	return wxEmptyString;
f3c82859 VS	160	}
5526e819	161
d1da8872	162	// NB: We convert input file to wchar_t here in Unicode mode, based on
2b5f62a0 VZ	163	// either Content-Type header or <meta> tags. In ANSI mode, we don't
	164	// do it as it is done by wxHtmlParser (for this reason, we add <meta>
	165	// tag if we used Content-Type header).
	166	#if wxUSE_UNICODE
d1da8872	167	int charsetPos;
2b5f62a0 VZ	168	if ((charsetPos = file.GetMimeType().Find(_T("; charset="))) != wxNOT_FOUND)
	169	{
	170	wxString charset = file.GetMimeType().Mid(charsetPos + 10);
	171	wxCSConv conv(charset);
	172	ReadString(doc, s, conv);
	173	}
	174	else
	175	{
	176	wxString tmpdoc;
	177	ReadString(tmpdoc, s, wxConvISO8859_1);
	178	wxString charset = wxHtmlParser::ExtractCharsetInformation(tmpdoc);
	179	if (charset.empty())
	180	doc = tmpdoc;
	181	else
	182	{
	183	wxCSConv conv(charset);
	184	doc = wxString(tmpdoc.mb_str(wxConvISO8859_1), conv);
	185	}
	186	}
	187	#else // !wxUSE_UNICODE
	188	ReadString(doc, s, wxConvLibc);
04dbb646	189	// add meta tag if we obtained this through http:
2b5f62a0	190	if (!file.GetMimeType().empty())
981e62aa	191	{
2b5f62a0 VZ	192	wxString hdr;
	193	wxString mime = file.GetMimeType();
	194	hdr.Printf(_T("<meta http-equiv=\"Content-Type\" content=\"%s\">"), mime.c_str());
	195	return hdr+doc;
981e62aa	196	}
2b5f62a0	197	#endif
981e62aa	198
3ca6a5f0	199	return doc;
5526e819 VS	200	}
	201
	202
	203
	204
	205	///// Module:
	206
	207	class wxHtmlFilterModule : public wxModule
	208	{
	209	DECLARE_DYNAMIC_CLASS(wxHtmlFilterModule)
	210
	211	public:
	212	virtual bool OnInit()
	213	{
	214	wxHtmlWindow::AddFilter(new wxHtmlFilterHTML);
	215	wxHtmlWindow::AddFilter(new wxHtmlFilterImage);
d1da8872	216	return true;
5526e819 VS	217	}
	218	virtual void OnExit() {}
	219	};
	220
	221	IMPLEMENT_DYNAMIC_CLASS(wxHtmlFilterModule, wxModule)
	222
8dd71e2b	223	#endif