//public static void TraceHtmlReader_v4(string file, bool generateCloseTag = false, bool disableLineColumn = false, bool disableScriptTreatment = false,
// bool useReadAttributeValue_v2 = true, bool useTranslateChar = true)
public static void TraceHtmlReader_v4(string file, HtmlReaderOptions options = HtmlReaderOptions.Default)
{
string traceFile = GetFile(file, ".trace.v4.txt");
Trace.WriteLine("HtmlReader v4");
Trace.WriteLine(" file \"{0}\"", file);
Trace.WriteLine(" trace \"{0}\"", traceFile);
//HtmlReader_v4.ReadFile(file, generateCloseTag: generateCloseTag, disableLineColumn: disableLineColumn, disableScriptTreatment: disableScriptTreatment,
// useReadAttributeValue_v2: useReadAttributeValue_v2, useTranslateChar: useTranslateChar)
// .zSave(traceFile);
HtmlReader_v4.ReadFile(file, options: options).zSave(traceFile);
}
//public static IEnumerable<HtmlNode> ReadString(string html, bool generateCloseTag = false, bool disableLineColumn = false,
// bool disableScriptTreatment = false, bool useReadAttributeValue_v2 = true, bool useTranslateChar = true)
public static IEnumerable <HtmlNode> ReadString(string html, HtmlReaderOptions options = HtmlReaderOptions.Default)
{
using (StringReader sr = new StringReader(html))
{
// attention return Read() generate exception Cannot read from a closed TextReader. (System.ObjectDisposedException)
//return Read(sr, generateCloseTag, disableLineColumn, disableScriptTreatment, useReadAttributeValue_v2, useTranslateChar);
//foreach (HtmlNode node in Read(sr, generateCloseTag, disableLineColumn, disableScriptTreatment, useReadAttributeValue_v2, useTranslateChar))
foreach (HtmlNode node in Read(sr, options))
{
yield return(node);
}
}
}
public static IEnumerable <HtmlDocNode> Read(TextReader textReader)
{
HtmlDocReader reader = new HtmlDocReader();
//bool disableLineColumn = false;
//bool disableScriptTreatment = false;
//bool useReadAttributeValue_v2 = true;
//bool useTranslateChar = true;
//reader._nodes = HtmlReader_v4.Read(textReader, generateCloseTag: true, disableLineColumn: disableLineColumn, disableScriptTreatment: disableScriptTreatment,
// useReadAttributeValue_v2: useReadAttributeValue_v2, useTranslateChar: useTranslateChar);
HtmlReaderOptions options = HtmlReaderOptions.Default | HtmlReaderOptions.GenerateCloseTag;
reader._nodes = HtmlReader_v4.Read(textReader, options);
return(reader.Read());
}
//HtmlReaderOptions
//public static IEnumerable<HtmlNode> Read(TextReader textReader, bool generateCloseTag = false, bool disableLineColumn = false,
// bool disableScriptTreatment = false, bool useReadAttributeValue_v2 = true, bool useTranslateChar = true, bool useFilterChar = true)
public static IEnumerable <HtmlNode> Read(TextReader textReader, HtmlReaderOptions options = HtmlReaderOptions.Default)
{
//HtmlReader_v4 htmlReader = new HtmlReader_v4(textReader, useTranslateChar, useFilterChar);
HtmlReader_v4 htmlReader = new HtmlReader_v4(textReader, options);
////htmlReader.GenerateCloseTag = generateCloseTag;
//htmlReader.GenerateCloseTag = (options & HtmlReaderOptions.GenerateCloseTag) == HtmlReaderOptions.GenerateCloseTag;
////htmlReader.DisableLineColumn = disableLineColumn;
//htmlReader.DisableLineColumn = (options & HtmlReaderOptions.DisableLineColumn) == HtmlReaderOptions.DisableLineColumn;
////htmlReader.DisableScriptTreatment = disableScriptTreatment;
//htmlReader.DisableScriptTreatment = (options & HtmlReaderOptions.DisableScriptTreatment) == HtmlReaderOptions.DisableScriptTreatment;
////htmlReader.UseReadAttributeValue_v2 = useReadAttributeValue_v2;
//htmlReader.TextReplaceControl = (options & HtmlReaderOptions.TextReplaceControl) == HtmlReaderOptions.TextReplaceControl;
return(htmlReader.Read());
}
//public HtmlReader_v4(TextReader textReader, bool useTranslateChar = true, bool useFilterChar = true)
public HtmlReader_v4(TextReader textReader, HtmlReaderOptions options = HtmlReaderOptions.Default)
{
_charStreamReader = new CharStreamReader(textReader);
//if (useTranslateChar)
if ((options & HtmlReaderOptions.UseTranslateChar) == HtmlReaderOptions.UseTranslateChar)
{
_charStreamReader.TranslateChar = TranslateChar;
}
//if (useFilterChar)
if ((options & HtmlReaderOptions.UseFilterChar) == HtmlReaderOptions.UseFilterChar)
{
_charStreamReader.FilterChar = FilterChar;
}
_generateCloseTag = (options & HtmlReaderOptions.GenerateCloseTag) == HtmlReaderOptions.GenerateCloseTag;
_disableLineColumn = (options & HtmlReaderOptions.DisableLineColumn) == HtmlReaderOptions.DisableLineColumn;
_disableScriptTreatment = (options & HtmlReaderOptions.DisableScriptTreatment) == HtmlReaderOptions.DisableScriptTreatment;
_textReplaceControl = (options & HtmlReaderOptions.TextReplaceControl) == HtmlReaderOptions.TextReplaceControl;
}
public static void FileHtmlToXml_v2(string file, string xmlFile, bool traceHtmlReader = false, bool traceHtmlToXml = false)
{
//HtmlReaderOptions options = HtmlReaderOptions.Default | HtmlReaderOptions.GenerateCloseTag;
HtmlReaderOptions options = HtmlReaderOptions.Default;
if (traceHtmlReader)
{
HtmlReader_v4.ReadFile(file, options: options | HtmlReaderOptions.DisableLineColumn).zSave(GetFile(file, ".HtmlReader.v4.txt"), jsonIndent: true);
}
using (StreamReader sr = zfile.OpenText(file))
{
HtmlReader_v4 htmlReader = new HtmlReader_v4(sr, options);
HtmlToXml_v3 htmlToXml = new HtmlToXml_v3(htmlReader);
htmlToXml.CreateXml().Save(xmlFile);
if (traceHtmlToXml)
{
htmlToXml.Log.zSave(GetFile(file, ".HtmlToXml.v2.json"));
}
}
}