/// <summary>
/// Načte aktuální uzel jako <see cref="XmlDocument"/>.
/// </summary>
/// <param name="reader"><see cref="XmlReader"/> na pozici elementu, jehož obsah se má načíst do dokumetnu.</param>
/// <returns>Vrací <see cref="XmlDocument"/> tvořený aktuálním uzlem nebo prázdný dokument.</returns>
public static XmlDocument ReadNodeAsXmlDocument(XmlReader reader)
{
XmlDocument xd = new XmlDocument();
if (reader.NodeType == XmlNodeType.Attribute)
{
reader.MoveToElement();
}
if (reader.NodeType != XmlNodeType.Element)
{
return(null);
}
string nodeName = reader.Name;
int depth = reader.Depth;
using (MemoryStream memoryStream = new MemoryStream())
{
XmlWriterSettings settings = new XmlWriterSettings();
settings.OmitXmlDeclaration = true;
using (XmlWriter xw = XmlWriter.Create(memoryStream, settings))
{
string text = null;
bool isEmpty = reader.IsEmptyElement;
try
{
Transformace.SerializeNode(reader, xw);
if (isEmpty && reader.Name == nodeName && reader.Depth == depth)
{
LoadXmlString(xw, memoryStream, xd);
return(xd);
}
while (reader.Read())
{
isEmpty = reader.IsEmptyElement;
Transformace.SerializeNode(reader, xw);
if (isEmpty && reader.Name == nodeName && reader.Depth == depth)
{
LoadXmlString(xw, memoryStream, xd);
break;
}
if (reader.NodeType == XmlNodeType.EndElement && reader.Name == nodeName && reader.Depth == depth)
{
LoadXmlString(xw, memoryStream, xd);
break;
}
}
}
catch (Exception e)
{
string error = e.Message;
}
}
}
return(xd);
}
public void VypisHeslaTypeUse()
{
if (base.VstupniSoubor == null || base.VystupniSoubor == null)
{
throw new ArgumentNullException("Nebyly zadány vhodné názvy vstupního nebo výstupního souboru.");
}
using (XmlReader r = Objekty.VytvorXmlReader(base.VstupniSoubor))
{
using (XmlWriter xw = Objekty.VytvorXmlWriter(base.VystupniSoubor))
{
xw.WriteStartDocument(true);
xw.WriteStartElement("entries");
while (r.Read())
{
if (r.NodeType == XmlNodeType.Element)
{
switch (r.Name)
{
case "entry":
string sUse = r.GetAttribute("use");
string sType = r.GetAttribute("type");
if (sUse == "public" && sType == "excl")
{
DPXT.SerializeNode(r, xw);
xw.WriteEndElement();
}
break;
default:
break;
}
}
}
xw.WriteEndElement();
}
}
}
/// <summary>
/// Upraví hranice heslové stati. Ohraničí jednotlivé části značkou <entry> a v rámci heslové stati seskupí významy podřazené značce <senseGrp>.
/// </summary>
public override void UpravitHraniceHesloveStati(string inputFile, string outputFile)
{
if (inputFile == null || outputFile == null)
{
throw new ArgumentNullException("Nebyly zadány vhodné názvy vstupního nebo výstupního souboru.");
}
using (XmlReader r = Objekty.VytvorXmlReader(inputFile))
{
using (XmlWriter xw = Objekty.VytvorXmlWriter(outputFile))
{
xw.WriteStartDocument(true);
bool blnPrvniEntryhead = true;
bool blnVyskytlySeVyznamy = false;
string strTypSenseGrp = null;
Stack <string> gstTypSenseGrp = new Stack <string>();
int intPocetOtevrenychSenseGrp = 0;
bool blnPrvniVyznam = false;
int xmlReaderDepp = 0;
int xmlWriterDeep = 0;
while (r.Read())
{
//Console.WriteLine("Reader: {0}, Writer: {1}", xmlReaderDepp, xmlWriterDeep);
if (r.NodeType == XmlNodeType.Element)
{
xmlReaderDepp++;
if (r.IsEmptyElement)
{
xmlReaderDepp--;
}
if (r.Name == "div1")
{
string name = r.Name;
}
switch (r.Name)
{
case "entryhead":
if (blnPrvniVyznam)
{
WriteEndElement(xw, ref xmlWriterDeep); //senses
xw.WriteWhitespace("\r\n");
}
blnPrvniVyznam = false;
blnVyskytlySeVyznamy = false;
int iPocetSenseGrp = gstTypSenseGrp.Count;
if (iPocetSenseGrp > 0)
{
for (int i = 1; i < iPocetSenseGrp; i++)
{
intPocetOtevrenychSenseGrp--;
WriteEndElement(xw, ref xmlWriterDeep); //senseGrp
xw.WriteWhitespace("\r\n");
}
gstTypSenseGrp.Clear();
}
while (intPocetOtevrenychSenseGrp > 0)
{
WriteEndElement(xw, ref xmlWriterDeep);
xw.WriteWhitespace("\r\n");
intPocetOtevrenychSenseGrp--;
}
if (!blnPrvniEntryhead)
{
WriteEndElement(xw, ref xmlWriterDeep); //entry
xw.WriteWhitespace("\r\n");
}
else
{
blnPrvniEntryhead = false;
}
WriteStartElement(xw, "entry", ref xmlWriterDeep);
xw.WriteWhitespace("\r\n");
goto default;
case "senseGrp":
strTypSenseGrp = r.GetAttribute("type");
intPocetOtevrenychSenseGrp++;
if (gstTypSenseGrp.Count > 0)
{
if (strTypSenseGrp == gstTypSenseGrp.Peek())
{
WriteEndElement(xw, ref xmlWriterDeep); //senseGrp
xw.WriteWhitespace("\r\n");
intPocetOtevrenychSenseGrp--;
}
else
{
if (blnVyskytlySeVyznamy)
{
WriteEndElement(xw, ref xmlWriterDeep);
xw.WriteWhitespace("\r\n");
intPocetOtevrenychSenseGrp--;
}
else
{
gstTypSenseGrp.Push(strTypSenseGrp);
}
}
}
else
{
gstTypSenseGrp.Push(strTypSenseGrp);
}
blnVyskytlySeVyznamy = false;
goto default;
case "sense":
if (!blnPrvniVyznam)
{
WriteStartElement(xw, "senses", ref xmlWriterDeep); //xw.WriteStartElement("senses");
xw.WriteWhitespace("\r\n");
blnPrvniVyznam = true;
}
blnVyskytlySeVyznamy = true;
goto default;
case "note":
case "appendix":
if (r.IsEmptyElement)
{
goto default;
}
while (gstTypSenseGrp.Count > 0)
{
//xw.WriteEndElement();
WriteEndElement(xw, ref xmlWriterDeep);
xw.WriteWhitespace("\r\n");
intPocetOtevrenychSenseGrp--;
gstTypSenseGrp.Pop();
}
if (blnPrvniVyznam)
{
//xw.WriteEndElement(); //senses
WriteEndElement(xw, ref xmlWriterDeep);
xw.WriteWhitespace("\r\n");
blnPrvniVyznam = false;
}
goto default;
default:
if (!r.IsEmptyElement)
{
xmlWriterDeep++;
}
DPXT.SerializeNode(r, xw);
break;
}
}
else
{
if (r.NodeType == XmlNodeType.EndElement)
{
xmlReaderDepp--;
switch (r.Name)
{
case "senseGrp":
if (blnPrvniVyznam)
{
//xw.WriteEndElement(); //senses
WriteEndElement(xw, ref xmlWriterDeep);
xw.WriteWhitespace("\r\n");
blnPrvniVyznam = false;
}
break;
case "div1":
//xw.WriteEndElement(); //entry
while (xmlWriterDeep > xmlReaderDepp + 1)
{
WriteEndElement(xw, ref xmlWriterDeep);
}
blnPrvniEntryhead = true;
blnPrvniVyznam = false;
goto default;
default:
xmlWriterDeep--;
DPXT.SerializeNode(r, xw);
break;
}
}
//else if (r.NodeType == XmlNodeType.XmlDeclaration && r.Name == "xml") {
// //nedělat nic, jde o začátek dokumentu; genruje cyhbu: nelze zapsat deklaraci, dokument již byl započat
//}
else
{
DPXT.SerializeNode(r, xw);
}
}
}
}
}
}
public void Run()
{
int id = 0;
XmlElementInfo currentElement = null;
XmlNamespaceManager namespaceManager = new XmlNamespaceManager(new NameTable());
namespaceManager.AddNamespace(String.Empty, NsTei);
XmlWriterSettings xmlWriterSettings = new XmlWriterSettings();
xmlWriterSettings.Indent = true;
List <string> ignoredElements = IgnoredElements ?? new List <string>();
XmlElementInfo ignoredElement = null;
string punctation = Punctation ?? String.Empty;
bool ignoreText = false;
bool?isRoot = null;
using (XmlReader xmlReader = XmlReader.Create(Input))
{
using (XmlWriter xmlWriter = XmlWriter.Create(Output, xmlWriterSettings))
{
xmlWriter.WriteStartDocument();
while (xmlReader.Read())
{
XmlNodeType nodeType = xmlReader.NodeType;
if (nodeType == XmlNodeType.Element)
{
if (!xmlReader.IsEmptyElement)
{
currentElement = XmlElementInfo.GetInfo(xmlReader, currentElement);
}
WriteXmlElementInfo(currentElement);
isRoot = !isRoot.HasValue;
if (ignoredElement == null && ignoredElements.Contains(xmlReader.LocalName))
{
ignoredElement = currentElement;
ignoreText = true;
}
}
if (nodeType == XmlNodeType.EndElement)
{
if (currentElement != null &&
ignoreText &&
currentElement.Name == ignoredElement.Name &&
currentElement.Depth == ignoredElement.Depth)
//if (ignoredElements.Contains(xmlReader.LocalName))
{
ignoredElement = null;
ignoreText = false;
}
if (currentElement != null)
{
currentElement = currentElement.Parent;
}
WriteXmlElementInfo(currentElement);
}
if (nodeType == XmlNodeType.Text)
{
if (ignoreText)
{
Transformace.SerializeNode(xmlReader, xmlWriter);
continue;
}
string text = xmlReader.Value;
if (text.IndexOf(' ') == -1)
{
xmlWriter.WriteElementString("w", NsTei, text);
}
else
{
_capacity = text.Length;
StringBuilder stringBuilder = new StringBuilder(_capacity);
foreach (char c in text)
{
if (c == ' ' || c == '\u0001' || c == '\t') //mezera a pevná mezera
{
WriteWordElement(ref stringBuilder, xmlWriter, _capacity, id++, XmlIdFormat);
xmlWriter.WriteStartElement("c");
xmlWriter.WriteAttributeString("type", "space");
xmlWriter.WriteAttributeString("space", NsXml, "preserve");
if (c == '\t')
{
xmlWriter.WriteString("\t");
}
else
{
xmlWriter.WriteString(" ");
}
xmlWriter.WriteEndElement();
}
else
{
if (punctation.IndexOf(c) > -1)
{
WriteWordElement(ref stringBuilder, xmlWriter, _capacity, id++, XmlIdFormat);
xmlWriter.WriteElementString("pc", NsTei, String.Format("{0}", c));
}
else
{
stringBuilder.Append(c);
}
}
}
//výpis posledního slova uloženého v bufferu
WriteWordElement(ref stringBuilder, xmlWriter, 0, id++, XmlIdFormat);
}
}
else
{
Transformace.SerializeNode(xmlReader, xmlWriter);
if (isRoot.HasValue && isRoot.Value)
{
xmlWriter.WriteAttributeString("xmlns", "nlp", null, NsNlp);
isRoot = false;
}
}
}
}
}
}
/// <summary>
/// Upraví hranice heslové stati, seskupí všechny prvky heslové stati do elementu <entry>. Využívá při tom značku <entryend>
/// </summary>
/// <exception cref="ArgumentNullException">Vyvolá výjimku, pokud nejsou zadány vstupní nebo výstupní soubor.</exception>
public override void UpravitHraniceHesloveStati(string inputFile, string outputFile)
{
//výchozí imnplementace se hodí pro ESSČ
string sChyba = null;
if (inputFile == null || outputFile == null)
{
throw new ArgumentNullException("Nebyly zadány vhodné názvy vstupního nebo výstupního souboru.");
}
using (XmlReader r = Objekty.VytvorXmlReader(inputFile))
{
using (XmlWriter xw = Objekty.VytvorXmlWriter(outputFile))
{
/*
* string strTypSenseGrp = null;
* Stack<string> gstTypSenseGrp = new Stack<string>();
* int intPocetOtevrenychSenseGrp = 0;
* bool blnVyskytlySeVyznamy = false;
*/
bool blnJeSenseGrp = false;
bool blnJeSense = false;
xw.WriteStartDocument(true);
bool blnPrvniEntryhead = true;
try
{
while (r.Read())
{
if (r.NodeType == XmlNodeType.Element)
{
switch (r.Name)
{
case "entryhead":
if (blnPrvniEntryhead)
{
xw.WriteStartElement("entry");
xw.WriteWhitespace("\r\n");
blnPrvniEntryhead = false;
}
DPXT.SerializeNode(r, xw);
blnJeSenseGrp = blnJeSense = false;
break;
case "entryend":
xw.WriteEndElement(); //entry
xw.WriteWhitespace("\r\n");
blnPrvniEntryhead = true;
break;
case "senseGrp":
if (blnJeSense)
{
xw.WriteEndElement(); //senses
}
if (blnJeSenseGrp)
{
xw.WriteEndElement(); //senseGrp
}
blnJeSenseGrp = true;
blnJeSense = false;
goto default;
case "sense":
if (!blnJeSense)
{
xw.WriteStartElement("senses");
}
blnJeSense = true;
goto default;
default:
if (blnJeSense && r.Depth == 2 && !(r.Name == "sense"))
{
xw.WriteEndElement(); //senses
blnJeSense = false;
}
if (blnJeSenseGrp && r.Depth == 2 && !r.Name.StartsWith("sense"))
{
blnJeSenseGrp = false;
xw.WriteEndElement(); //senseGrp
}
DPXT.SerializeNode(r, xw);
break;
}
}
else if (r.NodeType == XmlNodeType.EndElement)
{
if (r.Depth == 2)
{
switch (r.Name)
{
case "senseGrp":
/*
* if (blnJeSense)
* {
* xw.WriteEndElement(); //senseGrp
* blnJeSenseGrp = false;
* }
*/
//DPXT.SerializeNode(r, xw);
break;
case "sense":
DPXT.SerializeNode(r, xw);
break;
default:
if (blnJeSense)
{
xw.WriteEndElement(); //senses
blnJeSense = false;
}
DPXT.SerializeNode(r, xw);
break;
}
}
else
{
DPXT.SerializeNode(r, xw);
}
}
else
{
DPXT.SerializeNode(r, xw);
}
}
}
catch (Exception ex)
{
sChyba = ex.Message;
sChyba = r.ReadInnerXml();
while (sChyba.Trim().Length == 0)
{
if (r.Read())
{
sChyba = r.ReadInnerXml();
}
else
{
sChyba = "XmlReader je na konci souboru.";
}
}
Console.WriteLine(sChyba);
}
finally
{
if (xw.WriteState != WriteState.Error && xw.WriteState != WriteState.Closed)
{
xw.WriteEndDocument();
}
xw.Flush();
xw.Close();
}
}
}
}