public virtual bool SetXhtmlDocType(Node root)
{
string fpi = " ";
string sysid = "";
string namespace_Renamed = XHTML_NAMESPACE;
Node doctype;
doctype = root.FindDocType();
if (Options.DocType == TidyNet.DocType.Omit)
{
if (doctype != null)
Node.DiscardElement(doctype);
return true;
}
if (Options.DocType == TidyNet.DocType.Auto)
{
/* see what flavor of XHTML this document matches */
if ((this.versions & HtmlVersion.Html40Strict) != 0)
{
/* use XHTML strict */
fpi = "-//W3C//DTD XHTML 1.0 Strict//EN";
sysid = voyager_strict;
}
else if ((this.versions & HtmlVersion.Loose) != 0)
{
fpi = "-//W3C//DTD XHTML 1.0 Transitional//EN";
sysid = voyager_loose;
}
else if ((this.versions & HtmlVersion.Frames) != 0)
{
/* use XHTML frames */
fpi = "-//W3C//DTD XHTML 1.0 Frameset//EN";
sysid = voyager_frameset;
}
else
{
/* lets assume XHTML transitional */
fpi = "-//W3C//DTD XHTML 1.0 Transitional//EN";
sysid = voyager_loose;
}
}
else if (Options.DocType == TidyNet.DocType.Strict)
{
fpi = "-//W3C//DTD XHTML 1.0 Strict//EN";
sysid = voyager_strict;
}
else if (Options.DocType == TidyNet.DocType.Loose)
{
fpi = "-//W3C//DTD XHTML 1.0 Transitional//EN";
sysid = voyager_loose;
}
FixHtmlNameSpace(root, namespace_Renamed);
if (doctype == null)
{
doctype = NewNode(Node.DocTypeTag, this.lexbuf, 0, 0);
doctype.Next = root.Content;
doctype.Parent = root;
doctype.Prev = null;
root.Content = doctype;
}
if (Options.DocType == TidyNet.DocType.User && Options.DocTypeStr != null)
{
fpi = Options.DocTypeStr;
sysid = "";
}
this.txtstart = this.lexsize;
this.txtend = this.lexsize;
/* add public identifier */
AddStringLiteral("html PUBLIC ");
/* check if the fpi is quoted or not */
if (fpi[0] == '"')
{
AddStringLiteral(fpi);
}
else
{
AddStringLiteral("\"");
AddStringLiteral(fpi);
AddStringLiteral("\"");
}
if (sysid.Length + 6 >= this.Options.WrapLen)
{
AddStringLiteral("\n\"");
}
else
{
AddStringLiteral("\n \"");
}
/* add system identifier */
AddStringLiteral(sysid);
AddStringLiteral("\"");
this.txtend = this.lexsize;
doctype.Start = this.txtstart;
doctype.End = this.txtend;
return false;
}
/* fixup doctype if missing */
public virtual bool FixDocType(Node root)
{
Node doctype;
HtmlVersion guessed = HtmlVersion.Html40Strict;
int i;
if (this.badDoctype)
{
Report.Warning(this, null, null, Report.MALFORMED_DOCTYPE);
}
if (Options.XmlOut)
{
return true;
}
doctype = root.FindDocType();
if (Options.DocType == TidyNet.DocType.Omit)
{
if (doctype != null)
{
Node.DiscardElement(doctype);
}
return true;
}
if (Options.DocType == TidyNet.DocType.Strict)
{
Node.DiscardElement(doctype);
doctype = null;
guessed = HtmlVersion.Html40Strict;
}
else if (Options.DocType == TidyNet.DocType.Loose)
{
Node.DiscardElement(doctype);
doctype = null;
guessed = HtmlVersion.Html40Loose;
}
else if (Options.DocType == TidyNet.DocType.Auto)
{
if (doctype != null)
{
if (this.doctype == HtmlVersion.Unknown)
{
return false;
}
switch (this.doctype)
{
case HtmlVersion.Unknown:
return false;
case HtmlVersion.Html20:
if ((this.versions & HtmlVersion.Html20) != 0)
{
return true;
}
break; /* to replace old version by new */
case HtmlVersion.Html32:
if ((this.versions & HtmlVersion.Html32) != 0)
{
return true;
}
break; /* to replace old version by new */
case HtmlVersion.Html40Strict:
if ((this.versions & HtmlVersion.Html40Strict) != 0)
{
return true;
}
break; /* to replace old version by new */
case HtmlVersion.Html40Loose:
if ((this.versions & HtmlVersion.Html40Loose) != 0)
{
return true;
}
break; /* to replace old version by new */
case HtmlVersion.Frames:
if ((this.versions & HtmlVersion.Frames) != 0)
{
return true;
}
break; /* to replace old version by new */
}
/* INCONSISTENT_VERSION warning is now issued by ApparentVersion() */
}
/* choose new doctype */
guessed = GetHtmlVersion();
}
if (guessed == HtmlVersion.Unknown)
{
return false;
}
/* for XML use the Voyager system identifier */
if (this.Options.XmlOut || this.Options.XmlTags || this.isvoyager)
{
if (doctype != null)
Node.DiscardElement(doctype);
for (i = 0; i < W3CVersion.Length; ++i)
{
if (guessed == W3CVersion[i].Version)
{
FixHtmlNameSpace(root, W3CVersion[i].Profile);
break;
}
}
return true;
}
if (doctype == null)
{
doctype = NewNode(Node.DocTypeTag, this.lexbuf, 0, 0);
doctype.Next = root.Content;
doctype.Parent = root;
doctype.Prev = null;
root.Content = doctype;
}
this.txtstart = this.lexsize;
this.txtend = this.lexsize;
/* use the appropriate public identifier */
AddStringLiteral("html PUBLIC ");
if (Options.DocType == TidyNet.DocType.User && Options.DocTypeStr != null)
{
AddStringLiteral(Options.DocTypeStr);
}
else if (guessed == HtmlVersion.Html20)
{
AddStringLiteral("\"-//IETF//DTD HTML 2.0//EN\"");
}
else
{
AddStringLiteral("\"-//W3C//DTD ");
for (i = 0; i < W3CVersion.Length; ++i)
{
if (guessed == W3CVersion[i].Version)
{
AddStringLiteral(W3CVersion[i].Name);
break;
}
}
AddStringLiteral("//EN\"");
}
this.txtend = this.lexsize;
doctype.Start = this.txtstart;
doctype.End = this.txtend;
return true;
}
/// <summary> Internal routine that actually does the parsing. The caller
/// can pass either an InputStream or file name. If both are passed,
/// the file name is preferred.
/// </summary>
internal Node ParseInternal(Stream input, string file, Stream Output, TidyMessageCollection messages)
{
Lexer lexer;
Node document = null;
Node doctype;
Out o = new OutImpl(); /* normal output stream */
PPrint pprint;
/* ensure config is self-consistent */
_options.Adjust();
if (file != null)
{
input = new FileStream(file, FileMode.Open, FileAccess.Read);
}
else if (input == null)
{
input = Console.OpenStandardInput();
}
if (input != null)
{
lexer = new Lexer(new ClsStreamInImpl(input, _options.CharEncoding, _options.TabSize), _options);
lexer.messages = messages;
/*
* store pointer to lexer in input stream
* to allow character encoding errors to be
* reported
*/
lexer.input.Lexer = lexer;
/* Tidy doesn't alter the doctype for generic XML docs */
if (_options.XmlTags)
{
document = ParserImpl.parseXMLDocument(lexer);
}
else
{
document = ParserImpl.parseDocument(lexer);
if (!document.CheckNodeIntegrity())
{
Report.BadTree(lexer);
return(null);
}
Clean cleaner = new Clean(_options.tt);
/* simplifies <b><b> ... </b> ...</b> etc. */
cleaner.NestedEmphasis(document);
/* cleans up <dir>indented text</dir> etc. */
cleaner.List2BQ(document);
cleaner.BQ2Div(document);
/* replaces i by em and b by strong */
if (_options.LogicalEmphasis)
{
cleaner.EmFromI(document);
}
if (_options.Word2000 && cleaner.IsWord2000(document, _options.tt))
{
/* prune Word2000's <![if ...]> ... <![endif]> */
cleaner.DropSections(lexer, document);
/* drop style & class attributes and empty p, span elements */
cleaner.CleanWord2000(lexer, document);
}
/* replaces presentational markup by style rules */
if (_options.MakeClean || _options.DropFontTags)
{
cleaner.CleanTree(lexer, document);
}
if (!document.CheckNodeIntegrity())
{
Report.BadTree(lexer);
return(null);
}
doctype = document.FindDocType();
if (document.Content != null)
{
if (_options.Xhtml)
{
lexer.SetXhtmlDocType(document);
}
else
{
lexer.FixDocType(document);
}
if (_options.TidyMark)
{
lexer.AddGenerator(document);
}
}
/* ensure presence of initial <?XML version="1.0"?> */
if (_options.XmlOut && _options.XmlPi)
{
lexer.FixXmlPI(document);
}
if (document.Content != null)
{
Report.ReportVersion(lexer, doctype);
Report.ReportNumWarnings(lexer);
}
}
// Try to close the InputStream but only if if we created it.
if ((file != null) && (input != Console.OpenStandardOutput()))
{
try
{
input.Close();
}
catch (IOException)
{
}
}
if (lexer.messages.Errors > 0)
{
Report.NeedsAuthorIntervention(lexer);
}
o.State = StreamIn.FSM_ASCII;
o.Encoding = _options.CharEncoding;
if (lexer.messages.Errors == 0)
{
if (_options.BurstSlides)
{
Node body;
body = null;
/*
* remove doctype to avoid potential clash with
* markup introduced when bursting into slides
*/
/* discard the document type */
doctype = document.FindDocType();
if (doctype != null)
{
Node.DiscardElement(doctype);
}
/* slides use transitional features */
lexer.versions |= HtmlVersion.Html40Loose;
/* and patch up doctype to match */
if (_options.Xhtml)
{
lexer.SetXhtmlDocType(document);
}
else
{
lexer.FixDocType(document);
}
/* find the body element which may be implicit */
body = document.FindBody(_options.tt);
if (body != null)
{
pprint = new PPrint(_options);
Report.ReportNumberOfSlides(lexer, pprint.CountSlides(body));
pprint.CreateSlides(lexer, document);
}
else
{
Report.MissingBody(lexer);
}
}
else if (Output != null)
{
pprint = new PPrint(_options);
o.Output = Output;
if (_options.XmlTags)
{
pprint.PrintXmlTree(o, (short)0, 0, lexer, document);
}
else
{
pprint.PrintTree(o, (short)0, 0, lexer, document);
}
pprint.FlushLine(o, 0);
}
}
Report.ErrorSummary(lexer);
}
return(document);
}
