/// <summary>
/// main transform which needs the orginal File
/// </summary>
/// <param name="directionXSL">transform direction</param>
/// <param name="resourceResolver">xsl location</param>
/// <param name="originalFile">original File</param>
/// <param name="inputFile">File after pretreatment</param>
/// <param name="outputFile">output file</param>
protected void MainTransform(string directionXSL, XmlUrlResolver resourceResolver, string originalFile, string inputFile, string outputFile)
{
XPathDocument xslDoc;
XmlReaderSettings xrs = new XmlReaderSettings();
XmlReader source = null;
XmlWriter writer = null;
OoxZipResolver zipResolver = null;
string zipXMLFileName = "input.xml";
try
{
//xrs.ProhibitDtd = true;
xslDoc = new XPathDocument(((ResourceResolver)resourceResolver).GetInnerStream(directionXSL));
xrs.XmlResolver = resourceResolver;
string sr = ZipXMLFile(inputFile);
ZipReader archive = ZipFactory.OpenArchive(sr);
source = XmlReader.Create(archive.GetEntry(zipXMLFileName));
XslCompiledTransform xslt = new XslCompiledTransform();
XsltSettings settings = new XsltSettings(true, false);
xslt.Load(xslDoc, settings, resourceResolver);
if (!originalFile.Equals(string.Empty))
{
zipResolver = new OoxZipResolver(originalFile, resourceResolver);
}
XsltArgumentList parameters = new XsltArgumentList();
parameters.XsltMessageEncountered += new XsltMessageEncounteredEventHandler(MessageCallBack);
// zip format
parameters.AddParam("outputFile", "", outputFile);
// writer = new OoxZipWriter(inputFile);
writer = new UofZipWriter(outputFile);
if (zipResolver != null)
{
xslt.Transform(source, parameters, writer, zipResolver);
}
else
{
xslt.Transform(source, parameters, writer);
}
}
finally
{
if (writer != null)
{
writer.Close();
}
if (source != null)
{
source.Close();
}
}
}
public XmlContainer(BinaryReader _reader, uint size, uint typeCode, uint version, uint instance)
: base(_reader, size, typeCode, version, instance)
{
// Note: XmlContainers contain the data of a partial "unfinished"
// OOXML file (.zip based) as their body.
//
// I really don't like writing the data to a temp file just to
// be able to open it via ZipUtils.
//
// Possible alternatives:
// 1) Using System.IO.Compression -- supports inflation, but can't parse Zip header data
// 2) Modifying zlib + minizlib + ZipLib so I can pass in bytes, possible, but not worth the effort
string tempPath = Path.GetTempFileName();
try
{
using (FileStream fs = new FileStream(tempPath, FileMode.Create))
{
using (BinaryWriter tempStream = new BinaryWriter(fs))
{
int count = (int)this.Reader.BaseStream.Length;
byte[] bytes = this.Reader.ReadBytes(count);
tempStream.Write(bytes);
tempStream.Flush();
fs.Flush();
tempStream.Close();
fs.Close();
}
}
using (ZipReader zipReader = ZipFactory.OpenArchive(tempPath))
{
this.XmlDocumentElement = ExtractDocumentElement(zipReader, GetRelations(zipReader, ""));
}
}
finally
{
try
{
File.Delete(tempPath);
}
catch (IOException) { /* OK */ }
}
}
public XmlContainer(BinaryReader _reader, uint size, uint typeCode, uint version, uint instance)
: base(_reader, size, typeCode, version, instance)
{
// Note: XmlContainers contain the data of a partial "unfinished" OOXML file (.zip based) as their body.
//
// I really don't like writing the data to a temp file just to be able to open it via ZipUtils.
//
// Possible alternatives:
// 1) Using System.IO.Compression -- supports inflation, but can't parse Zip header data
// 2) Modifying zlib + minizlib + ZipLib so I can pass in bytes, possible, but not worth the effort
// KH - I've left the original comment above, but I've ported this to use option (1) as the IZipLib result can't read headers anyway - it can only open entries.
using (var zipReader = ZipFactory.OpenArchive(this.Reader.BaseStream))
this.XmlDocumentElement = ExtractDocumentElement(zipReader, GetRelations(zipReader, ""));
}
private void DoUofToOoxTransform(string inputFile, string outputFile, string resourceDir)
{
guid = Guid.NewGuid();
if (!IsUof(inputFile))
{
throw new NotAnUofDocumentException(inputFile + " 不是UOF文件");
}
LinkedList <IProcessor> .Enumerator enumer = uof2ooxPreProcessors.GetEnumerator();
string outputPath = Path.GetTempPath().ToString() + guid.ToString();
string output = outputPath + Path.DirectorySeparatorChar + "tmpDoc";
if (!Directory.Exists(outputPath))
{
Directory.CreateDirectory(outputPath);
}
// string input = EIPicturePre(inputFile, outputPath);
ZipReader archive = ZipFactory.OpenArchive(inputFile);
archive.ExtractUOFDocument(inputFile, outputPath);
string input = inputFile;
int i = 0;
while (enumer.MoveNext())
{
i++;
enumer.Current.OriginalFilename = inputFile;
enumer.Current.InputFilename = input;
enumer.Current.OutputFilename = output + i + ".xml";
enumer.Current.transform();
input = enumer.Current.OutputFilename;
}
DoUofToOoxMainTransform(enumer.Current.OutputFilename, outputFile, resourceDir);
// Directory.Delete(outputPath, true);
}
public override bool transform()
{
FileStream fs = null;
XmlTextReader txtReader = null;
string extractPath = Path.GetDirectoryName(outputFile) + "\\";
ZipReader archive = ZipFactory.OpenArchive(inputFile);
archive.ExtractOfficeDocument(inputFile, extractPath);
string prefix = this.GetType().Namespace + "." + TranslatorConstants.RESOURCE_LOCATION + ".Powerpoint.oox2uof";//路径
Assembly asm = Assembly.GetExecutingAssembly();
foreach (string name in asm.GetManifestResourceNames())
{
if (name.StartsWith(prefix))
{
string filename = name.Substring(prefix.Length + 1);
FileStream writer = new FileStream(extractPath + filename, FileMode.Create);
Stream baseStream = asm.GetManifestResourceStream(name);
int Length = 10240;
Byte[] buffer = new Byte[Length];
int bytesRead = baseStream.Read(buffer, 0, Length);
while (bytesRead > 0)
{
writer.Write(buffer, 0, bytesRead);
bytesRead = baseStream.Read(buffer, 0, Length);
}
baseStream.Close();
writer.Close();
}
}
bool isSuccess = true;
try
{
// add theme rels
AddThemeRels(extractPath);
File.Copy(extractPath + "pre1.xsl", extractPath + @"\ppt\pre1.xsl", true);
XPathDocument doc = new XPathDocument(extractPath + @"\ppt\presentation.xml");
XslCompiledTransform transFrom = new XslCompiledTransform();
XsltSettings setting = new XsltSettings(true, false);
XmlUrlResolver xur = new XmlUrlResolver();
transFrom.Load(extractPath + @"\ppt\pre1.xsl", setting, xur);
XPathNavigator nav = ((IXPathNavigable)doc).CreateNavigator();
fs = new FileStream(extractPath + "pre1tmp.xml", FileMode.Create);
transFrom.Transform(nav, null, fs);
fs.Close();
doc = new XPathDocument(extractPath + "pre1tmp.xml");
nav = ((IXPathNavigable)doc).CreateNavigator();
fs = new FileStream(extractPath + "pre2tmp.xml", FileMode.Create);
Assembly ass = Assembly.Load("ppt_oox2uof");//使用预编译后的xslt
Type t = ass.GetType("pre2");
transFrom.Load(t);
transFrom.Transform(nav, null, fs);
fs.Close();
xmlDoc = new XmlDocument();
txtReader = new XmlTextReader(extractPath + "pre2tmp.xml");
xmlDoc.Load(txtReader);
txtReader.Close();
setNSManager(); //
string currentPath = "//p:sp";
purifyMethod(currentPath); //
//2011-01-12罗文甜：阴影预处理
XmlNodeList shapeShadeList = xmlDoc.SelectNodes("//a:outerShdw", nm);
if (shapeShadeList != null)
{
foreach (XmlNode shapeShade in shapeShadeList)
{
double dist, dir;
if (((XmlElement)shapeShade).HasAttribute("dist"))
{
dist = Convert.ToDouble(shapeShade.Attributes.GetNamedItem("dist").Value) / 12700;
}
else
{
dist = 0.0;
}
if (((XmlElement)shapeShade).HasAttribute("dir"))
{
dir = (Convert.ToDouble(shapeShade.Attributes.GetNamedItem("dir").Value) * 3.1415) / (180 * 60000);
}
else
{
dir = 0.0;
}
double xValue = dist * Math.Cos(dir);
double yValue = dist * Math.Sin(dir);
XmlElement x = xmlDoc.CreateElement("x");
x.InnerText = Convert.ToString(xValue);
shapeShade.AppendChild(x);
XmlElement y = xmlDoc.CreateElement("y");
y.InnerText = Convert.ToString(yValue);
shapeShade.AppendChild(y);
}
}
resultWriter = new XmlTextWriter(outputFile, System.Text.Encoding.UTF8);
xmlDoc.Save(resultWriter);
resultWriter.Close();
string tmpFile = extractPath + "tmpID.xml";
ChangeIDVal(outputFile, tmpFile);//
}
catch (Exception e)
{
logger.Error("Fail in OoxToUofPreProcessorOnePresentation: " + e.Message);
logger.Error(e.StackTrace);
isSuccess = false;
throw new Exception("Fail in OoxToUofPreProcessorOnePresentation");
}
finally
{
if (resultWriter != null)
{
resultWriter.Close();
}
if (fs != null)
{
fs.Close();
}
if (txtReader != null)
{
txtReader.Close();
}
}
return(isSuccess);
}
///// <summary>
///// Initialize according to the input file type
///// </summary>
///// <param name="inputFileType">DocType of input file</param>
///// <returns>return IUOFTranslator</returns>
//public static IUOFTranslator CheckFileType(DocType inputFileType)
//{
// switch (inputFileType)
// {
// case DocType.Word: return new WordTranslator();
// case DocType.Powerpoint: return new PresentationTranslator();
// case DocType.Excel: return new SpreadsheetTranslator();
// default: throw new Exception("not an office 2010 file");
// }
//}
/// <summary>
/// check Microsoft file type
/// </summary>
/// <param name="srcFileName">source file name</param>
/// <returns>document type</returns>
public static MSDocType CheckMSFileType(string srcFileName)
{
FileInfo fi = new FileInfo(srcFileName);
if (File.Exists(srcFileName))
{
XmlReader source = null;
ZipReader archive = ZipFactory.OpenArchive(srcFileName);
// get the main entry xml file
string entry = string.Empty;
switch (fi.Extension.ToLower())
{
case ".docx": entry = TranslatorMgrConstants.WordDocument_xml; break;
case ".pptx": entry = TranslatorMgrConstants.PresentationDocument_xml; break;
case ".xlsx": entry = TranslatorMgrConstants.SpreadsheetDocument_xml; break;
}
source = XmlReader.Create(archive.GetEntry(entry));
XmlDocument xdoc = new XmlDocument();
xdoc.Load(source);
// namespace of strict document
XmlNamespaceManager nm = new XmlNamespaceManager(xdoc.NameTable);
nm.AddNamespace("w", TranslatorMgrConstants.XMLNS_W);
nm.AddNamespace("p", TranslatorMgrConstants.XMLNS_P);
nm.AddNamespace("ws", TranslatorMgrConstants.XMLNS_WS);
switch (fi.Extension.ToLower())
{
case ".docx":
{
// @w:conformance
if (xdoc.SelectSingleNode("w:document", nm) != null)
{
return(MSDocType.StrictWord);
}
else
{
return(MSDocType.TransitionalWord);
}
}
case ".pptx":
{
if (xdoc.SelectSingleNode("p:presentation", nm) != null)
{
return(MSDocType.StrictPowerpoint);
}
else
{
return(MSDocType.TransitionalPowerpoint);
}
}
case ".xlsx":
{
if (xdoc.SelectSingleNode("ws:workbook", nm) != null)
{
return(MSDocType.StrictExcel);
}
else
{
return(MSDocType.TransitionalExcel);
}
}
default: return(MSDocType.Unknown);
}
}
else
{
return(MSDocType.Unknown);
}
}
/// <summary>
/// Constructor
/// </summary>
/// <param name="filename">Absolute path to the Zip package</param>
public ZipResolver(String filename)
{
archive = ZipFactory.OpenArchive(filename);
// initialize hash table of odf resource file names
entries = new Hashtable();
}
protected override void DoOoxToUofMainTransform(string originalFile, string inputFile, string outputFile, string resourceDir)
{
string wordPrePath = Path.GetDirectoryName(inputFile) + Path.DirectorySeparatorChar;
string grpTmp = Path.GetDirectoryName(inputFile) + "\\" + "grpTmp.xml";//把grpTmp.xml放在temp下的缓存文件夹下
string drawingRelTmp = Path.GetDirectoryName(inputFile) + "\\" + "drawingRelTmp.xml";
string picture_xml = "ppt\\media";
string mediaPath = wordPrePath + picture_xml;
XmlDocument xdoc = new XmlDocument();
xdoc.Load(inputFile);//tempdoc1.xml
XmlNamespaceManager nm = new XmlNamespaceManager(xdoc.NameTable);
nm.AddNamespace("p", TranslatorConstants.XMLNS_P);
nm.AddNamespace("a", TranslatorConstants.XMLNS_A);
nm.AddNamespace("w", TranslatorConstants.XMLNS_W);
nm.AddNamespace("dsp", TranslatorConstants.XMLNS_DSP);
FileStream fs = new FileStream(grpTmp, FileMode.Create);
try
{
GrpShPre(xdoc, nm, "p", fs);
}
catch (Exception ex)
{
throw ex;
}
finally
{
if (fs != null)
{
fs.Close();
}
}
// chart转换成图片
string charts_xml = wordPrePath + "ppt\\charts";
if (Directory.Exists(charts_xml))
{
string[] charts = Directory.GetFiles(charts_xml, "chart*");
if (charts.Length > 0)
{
if (!Directory.Exists(mediaPath))
{
Directory.CreateDirectory(mediaPath);
}
}
foreach (string chartXMLFile in charts)
{
SaveChartAsPic(chartXMLFile, mediaPath + Path.AltDirectorySeparatorChar + Path.GetFileName(chartXMLFile) + ".jpg", System.Windows.Forms.DataVisualization.Charting.ChartImageFormat.Jpeg);
}
}
//ole对象处理lyy
string tmpPic = wordPrePath + "tmpOle.xml";
if (Directory.Exists(wordPrePath + "ppt\\embeddings"))
{
xdoc.Load(grpTmp);
xdoc = OlePretreatment(xdoc, "p:presentation", wordPrePath, nm);
xdoc.Save(tmpPic);
grpTmp = tmpPic;
}
// 图片预处理
tmpPic = wordPrePath + "tmpPic.xml";
if (Directory.Exists(mediaPath))
{
xdoc.Load(grpTmp);
xdoc = PicPretreatment(xdoc, "p:presentation", mediaPath, nm);
xdoc.Save(tmpPic);
grpTmp = tmpPic;
}
//ole对象处理
AddDrawingRel(xdoc, nm, drawingRelTmp);
XmlReaderSettings xrs = new XmlReaderSettings();
XmlReader source = null;
string toMainSheet = ZipXMLFile(drawingRelTmp);
string zipXMLFileName = "input.xml";
//string sr = ZipXMLFile(inputFile);
ZipReader archive = ZipFactory.OpenArchive(toMainSheet);
source = XmlReader.Create(archive.GetEntry(zipXMLFileName));
//xdoc.Load(drawingRelTmp);
XslCompiledTransform transFrom = new XslCompiledTransform();
XsltSettings setting = new XsltSettings(true, false);
XmlUrlResolver xur = new XmlUrlResolver();
XPathNavigator nav = ((IXPathNavigable)xdoc).CreateNavigator();//root
fs = null;
string mainOutputFile = Path.GetDirectoryName(inputFile) + "\\" + "mainOutputFile.xml";//缓存文件中保存mainOutputFile.xml
fs = new FileStream(mainOutputFile, FileMode.Create);
string mainSheet = Path.GetDirectoryName(inputFile).ToString() + "\\" + TranslatorConstants.OOXToUOF_XSL;//从缓存中调用oox2uof.xsl
transFrom.Load(mainSheet, setting, xur);
//Assembly ass = Assembly.Load("ppt_oox2uof");//调用ppt_oox2uof.dll程序集
//Type t = ass.GetType("oox2uof");//name=oox2uof.xslt
//transFrom.Load(t);
//transFrom.Transform(nav, null, fs);
transFrom.Transform(source, null, fs);
fs.Close();
mainOutput = mainOutputFile;
//XmlUrlResolver resourceResolver = null;
//try
//{
// resourceResolver = new ResourceResolver(Assembly.GetExecutingAssembly(),
// this.GetType().Namespace + "." + TranslatorConstants.RESOURCE_LOCATION + "." + "Powerpoint.oox2uof");
// MainTransform(TranslatorConstants.OOXToUOF_XSL, resourceResolver, originalFile, grpTmp, outputFile);
////}
//catch (Exception ex)
//{
// logger.Warn(ex.Message);
//}
}
public OoxZipResolver(String filename, XmlUrlResolver resourceResolver)
{
archive = ZipFactory.OpenArchive(filename);
entries = new Hashtable();
this.resourceResolver = resourceResolver;
}
/// <summary>
/// Check the validity of an Office Open XML file.
/// </summary>
/// <param name="fileName">The path of the docx file.</param>
public void validate(String fileName)
{
// 0. The file must exist and be a valid Zip archive
try
{
reader = ZipFactory.OpenArchive(fileName);
}
catch (ZipException e)
{
throw new NotAnOoxDocumentException("Problem opening the pptx file : " + e.Message);
}
catch (Exception e)
{
throw new PptxValidatorException("Problem opening the pptx file : " + e.Message);
}
// 1. [Content_Types].xml must be present and valid
Stream contentTypes = null;
try
{
contentTypes = reader.GetEntry(OOX_CONTENT_TYPE_FILE);
}
catch (Exception)
{
//modified by lohith - to display user friendly message
throw new NotAnOoxDocumentException("The pptx package must have a \"/[Content_Types].xml\" file");
//throw new PptxValidatorException("The pptx package must have a \"/[Content_Types].xml\" file");
}
this.validateXml(contentTypes);
// Code to verify whether a file is valid pptx file or not
//bug no. 1698280
//Code chcanges 1 of 2
if (!this.validatePptx(reader))
{
throw new NotAnOoxDocumentException("The pptx package must have a Presentation.xml file");
}
// End
// 2. _rels/.rels must be present and valid
Stream relationShips = null;
try
{
relationShips = reader.GetEntry(OOX_RELATIONSHIP_FILE);
}
catch (Exception)
{
//modified by lohith - to display user friendly message
throw new NotAnOoxDocumentException("The pptx package must have a a \"/_rels/.rels\" file");
//throw new PptxValidatorException("The pptx package must have a \"/_rels/.rels\" file");
}
this.validateXml(relationShips);
// 3. _rel/.rels must contain a relationship of type openDocument
relationShips = reader.GetEntry(OOX_RELATIONSHIP_FILE);
XmlReader r = XmlReader.Create(relationShips);
String docTarget = null;
while (r.Read() && docTarget == null)
{
if (r.NodeType == XmlNodeType.Element && r.GetAttribute("Type") == OOX_DOCUMENT_RELATIONSHIP_TYPE)
{
docTarget = r.GetAttribute("Target");
}
}
if (docTarget == null)
{
//modified by lohith - to display user friendly message
throw new NotAnOoxDocumentException("openDocument relation not found in \"/_rels/.rels\"");
//throw new PptxValidatorException("openDocument relation not found in \"/_rels/.rels\"");
}
// 4. For each item in _rels/.rels
relationShips = reader.GetEntry(OOX_RELATIONSHIP_FILE);
r = XmlReader.Create(relationShips);
while (r.Read())
{
if (r.NodeType == XmlNodeType.Element && r.LocalName == "Relationship")
{
String target = r.GetAttribute("Target");
// 4.1. The target item must exist in the package
Stream item = null;
try
{
item = reader.GetEntry(target);
}
catch (Exception)
{
//modified by lohith - to display user friendly message
throw new NotAnOoxDocumentException("The file \"" + target + "\" is described in the \"/_rels/.rels\" file but does not exist in the package.");
//throw new PptxValidatorException("The file \"" + target + "\" is described in the \"/_rels/.rels\" file but does not exist in the package.");
}
// 4.2. A content type can be found in [Content_Types].xml file
String ct = this.findContentType(reader, "/" + target);
// 4.3. If it's an xml file, it has to be valid
if (ct.EndsWith("+xml"))
{
this.validateXml(item);
}
}
}
Stream partRel = null;
String partDir = docTarget.Substring(0, docTarget.IndexOf("/"));
String partRelPath = partDir + "/_rels/" + docTarget.Substring(docTarget.IndexOf("/") + 1) + ".rels";
bool partRelExists = true;
try
{
partRel = reader.GetEntry(partRelPath);
}
catch (Exception)
{
partRelExists = false;
}
if (partRelExists)
{
this.validateXml(partRel);
// 5. For each item in /ppt/_rels/presentation.xml.rels
partRel = reader.GetEntry(partRelPath);
r = XmlReader.Create(partRel);
ValidateRels(partDir, r);
//retrieve all ids referenced in the document
r.Close();
Stream doc = reader.GetEntry(docTarget);
r = XmlReader.Create(doc);
ArrayList ids = new ArrayList();
while (r.Read())
{
if (r.LocalName == "custShow")
{
r.Skip();
}
else
{
if (r.NodeType == XmlNodeType.Element && r.GetAttribute("id", OOX_DOC_REL_NS) != null)
{
ids.Add(r.GetAttribute("id", OOX_DOC_REL_NS));
}
}
}
// check if each id exists in the partRel file
if (ids.Count != 0)
{
if (!partRelExists)
{
//modified by lohith - to display user friendly message
throw new NotAnOoxDocumentException("Referenced id exist but no part relationship file found");
//throw new PptxValidatorException("Referenced id exist but no part relationship file found");
}
relationShips = reader.GetEntry(partRelPath);
r = XmlReader.Create(relationShips);
while (r.Read())
{
if (r.NodeType == XmlNodeType.Element && r.LocalName == "Relationship")
{
if (ids.Contains(r.GetAttribute("Id")))
{
ids.Remove(r.GetAttribute("Id"));
}
}
}
if (ids.Count != 0)
{
//modified by lohith - to display user friendly message
throw new NotAnOoxDocumentException("One or more relationship id have not been found in the partRelationship file : " + ids[0]);
//throw new PptxValidatorException("One or more relationship id have not been found in the partRelationship file : " + ids[0]);
}
}
}
//6. For each item in /ppt/slideMasters/_rels/slideMasters.xml.rels
docTarget = null;
partRel = reader.GetEntry(partRelPath);
r = XmlReader.Create(partRel);
while (r.Read() && docTarget == null)
{
if (r.NodeType == XmlNodeType.Element && r.GetAttribute("Type") == OOX_SIDEMASTER_RELATIONSHIP_TYPE)
{
docTarget = r.GetAttribute("Target");
partDir = OOX_SIDEMASTER_RELATIONSHIP_FILE;
partRelPath = partDir + "/_rels/" + docTarget.Substring(docTarget.IndexOf("/") + 1) + ".rels";
partRel = reader.GetEntry(partRelPath);
r = XmlReader.Create(partRel);
ValidateRels(partDir, r);
}
}
//7. Validation for SlideLayouts if required will be added here.
//8. For each slide in a presentation, validation needs to be added here
//9. During developement required validations for pptx feature e.g notes master, numbering etc. will be added.
//10. are all referenced relationships exist in the part relationship file
}
public override bool transform()
{
//string Document_xml = "word/document.xml";//word下的doucument.xml文档
string picture_xml = "word\\media";
// string tmpFile = Path.GetTempPath() + "tmpDoc.xml";
// string tmpFile2 = Path.GetTempPath() + "sndtmpDoc.xml";
// Guid gPath = Guid.NewGuid();
// string wordPrePath = Path.GetTempPath() + gPath.ToString() + "\\";
// if (!Directory.Exists(wordPrePath))
// Directory.CreateDirectory(wordPrePath);
string wordPrePath = Path.GetDirectoryName(outputFile) + "\\";//输出文档路径
string mediaPath = wordPrePath + picture_xml;
string tblVtcl = wordPrePath + "tblVertical.xml";
string tmpFile = wordPrePath + "tmpDoc.xml";
string tmpFile2 = wordPrePath + "sndtmpDoc.xml"; //预处理中的中间文档
string tmpFile3 = wordPrePath + "thrtmpDoc.xml"; //预处理中的中间文档
string preOutputFileName = wordPrePath + "tempdoc.xml"; //预处理出来的中间文档
XmlReader source = null;
XmlWriter writer = null;
ZipReader archive = ZipFactory.OpenArchive(inputFile);
archive.ExtractOfficeDocument(inputFile, wordPrePath);
//archive.ExtractUOFDocument(inputFile, wordPrePath);
bool isSuccess = true;
try
{
OoxToUofTableProcessing tableProcessing2 = new OoxToUofTableProcessing(wordPrePath + "word" + Path.AltDirectorySeparatorChar + "document.xml");
tableProcessing2.Processing(tblVtcl);
//找到预处理第一步的式样单pretreatmentStep1.xsl
XPathDocument xpdoc = UOFTranslator.GetXPathDoc(TranslatorConstants.OOXToUOF_PRETREAT_STEP1_XSL, TranslatorConstants.OOXToUOF_WORD_LOCATION);
XslCompiledTransform xslt = new XslCompiledTransform();
xslt.Load(xpdoc);
// source = XmlReader.Create(archive.GetEntry(Document_xml));//解压，得到document.xml文档
source = XmlReader.Create(tblVtcl);
writer = new XmlTextWriter(tmpFile, Encoding.UTF8); //tmpDoc.xml
xslt.Transform(source, writer); //document.xml--经过预处理.xsl--tmpDoc.xml
if (writer != null)
{
writer.Close();
}
if (source != null)
{
source.Close();
}
//第二步预处理
xpdoc = UOFTranslator.GetXPathDoc(TranslatorConstants.OOXToUOF_PRETREAT_STEP2_XSL, TranslatorConstants.OOXToUOF_WORD_LOCATION);
xslt = new XslCompiledTransform();
xslt.Load(xpdoc);
writer = new XmlTextWriter(tmpFile2, Encoding.UTF8); //sndtemDoc.xml
xslt.Transform(tmpFile, writer); //tmpDoc--经过预处理2.xsl--sndtemDoc.xml
if (writer != null)
{
writer.Close();
}
//第三步预处理
xpdoc = UOFTranslator.GetXPathDoc(TranslatorConstants.OOXToUOF_PRETREAT_STEP3_XSL, TranslatorConstants.OOXToUOF_WORD_LOCATION);
xslt = new XslCompiledTransform();
xslt.Load(xpdoc);
// xslt.Transform(tmpFile2, outputFile);
xslt.Transform(tmpFile2, tmpFile3);//经过预处理3--tempdoc.xml
//2011/6/17zhaobj：阴影预处理
XmlNamespaceManager nms = null;
XmlDocument xmlDoc;
XmlTextWriter resultWriter = null;
xmlDoc = new XmlDocument();
xmlDoc.Load(tmpFile3);
nms = new XmlNamespaceManager(xmlDoc.NameTable);
nms.AddNamespace("a", TranslatorConstants.XMLNS_A);
nms.AddNamespace("w", TranslatorConstants.XMLNS_W);
XmlNodeList shapeShadeList = xmlDoc.SelectNodes("//a:outerShdw", nms);
if (shapeShadeList != null)
{
foreach (XmlNode shapeShade in shapeShadeList)
{
double dist, dir;
if (((XmlElement)shapeShade).HasAttribute("dist"))
{
dist = Convert.ToDouble(shapeShade.Attributes.GetNamedItem("dist").Value) / 12700;
}
else
{
dist = 0.0;
}
if (((XmlElement)shapeShade).HasAttribute("dir"))
{
dir = (Convert.ToDouble(shapeShade.Attributes.GetNamedItem("dir").Value) * 3.1415) / (180 * 60000);
}
else
{
dir = 0.0;
}
double xValue = dist * Math.Cos(dir);
double yValue = dist * Math.Sin(dir);
XmlElement x = xmlDoc.CreateElement("x");
x.InnerText = Convert.ToString(xValue);
shapeShade.AppendChild(x);
XmlElement y = xmlDoc.CreateElement("y");
y.InnerText = Convert.ToString(yValue);
shapeShade.AppendChild(y);
}
}
resultWriter = new XmlTextWriter(preOutputFileName, System.Text.Encoding.UTF8);
xmlDoc.Save(resultWriter);
resultWriter.Close();
/*
*
*
*
*/
OutputFilename = preOutputFileName;
//图片预处理
//if (Directory.Exists(wordPrePath + picture_xml))
//{
// xmlDoc.Load(preOutputFileName);
// DirectoryInfo mediaInfo = new DirectoryInfo(wordPrePath + picture_xml);
// FileInfo[] medias = mediaInfo.GetFiles();
// XmlNode root = xmlDoc.SelectSingleNode("w:document", nms);
// XmlElement mediaNode=xmlDoc.CreateElement("w","media",TranslatorConstants.XMLNS_W);
// foreach (FileInfo media in medias)
// {
// XmlElement mediaFileNode = xmlDoc.CreateElement("u2opic", "picture", "urn:u2opic:xmlns:post-processings:special");
// mediaFileNode.SetAttribute("target", "urn:u2opic:xmlns:post-processings:special", media.FullName);
// mediaNode.AppendChild(mediaFileNode);
// }
// root.AppendChild(mediaNode);
// xmlDoc.Save(tmpPic);
// OutputFilename = tmpPic;
//}
}
catch (Exception e)
{
logger.Error("Fail in OoxToUofPreProcessorOneWord: " + e.Message);
logger.Error(e.StackTrace);
isSuccess = false;
}
finally
{
if (writer != null)
{
writer.Close();
}
if (source != null)
{
source.Close();
}
if (archive != null)
{
archive.Close();
}
if (File.Exists(tmpFile2))
{
File.Delete((tmpFile2));
}
}
return(isSuccess);
}
/// <summary>
/// Check the validity of an Office Open XML file.
/// </summary>
/// <param name="fileName">The path of the docx file.</param>
public void validate(string fileName)
{
// 0. The file must exist and be a valid Zip archive
ZipReader reader = null;
try
{
reader = ZipFactory.OpenArchive(fileName);
}
catch (Exception e)
{
throw new ValidationException("Problem opening the docx file : " + e.Message);
}
// 1. [Content_Types].xml must be present and valid
Stream contentTypes = null;
try
{
contentTypes = reader.GetEntry(OOX_CONTENT_TYPE_FILE);
}
catch (Exception)
{
throw new ValidationException("The docx package must have a \"/[Content_Types].xml\" file");
}
this.validateXml(contentTypes);
// 2. _rels/.rels must be present and valid
Stream relationShips = null;
try
{
relationShips = reader.GetEntry(OOX_RELATIONSHIP_FILE);
}
catch (Exception)
{
throw new ValidationException("The docx package must have a \"/_rels/.rels\" file");
}
this.validateXml(relationShips);
// 3. _rel/.rels must contain a relationship of type openDocument
relationShips = reader.GetEntry(OOX_RELATIONSHIP_FILE);
XmlReader r = XmlReader.Create(relationShips);
String docTarget = null;
while (r.Read() && docTarget == null)
{
if (r.NodeType == XmlNodeType.Element && r.GetAttribute("Type") == OOX_DOCUMENT_RELATIONSHIP_TYPE)
{
docTarget = r.GetAttribute("Target");
}
}
if (docTarget == null)
{
throw new ValidationException("openDocument relation not found in \"/_rels/.rels\"");
}
// 4. For each item in _rels/.rels
relationShips = reader.GetEntry(OOX_RELATIONSHIP_FILE);
r = XmlReader.Create(relationShips);
while (r.Read())
{
if (r.NodeType == XmlNodeType.Element && r.LocalName == "Relationship")
{
String target = r.GetAttribute("Target");
// 4.1. The target item must exist in the package
Stream item = null;
try
{
item = reader.GetEntry(target);
}
catch (Exception)
{
throw new ValidationException("The file \"" + target + "\" is described in the \"/_rels/.rels\" file but does not exist in the package.");
}
// 4.2. A content type can be found in [Content_Types].xml file
String ct = this.findContentType(reader, "/" + target);
// 4.3. If it's an xml file, it has to be valid
if (ct.EndsWith("+xml"))
{
this.validateXml(item);
}
}
}
// Does a part relationship exist ?
Stream partRel = null;
String partDir = docTarget.Substring(0, docTarget.IndexOf("/"));
String partRelPath = partDir + "/_rels/" + docTarget.Substring(docTarget.IndexOf("/") + 1) + ".rels";
bool partRelExists = true;
try
{
partRel = reader.GetEntry(partRelPath);
}
catch (Exception)
{
partRelExists = false;
}
if (partRelExists)
{
this.validateXml(partRel);
// 5. For each item in /word/_rels/document.xml.rels
partRel = reader.GetEntry(partRelPath);
r = XmlReader.Create(partRel);
while (r.Read())
{
if (r.NodeType == XmlNodeType.Element && r.LocalName == "Relationship")
{
String target = partDir + "/" + r.GetAttribute("Target");
// Is the target item exist in the package ?
Stream item = null;
bool fileExists = true;
try
{
item = reader.GetEntry(target);
}
catch (Exception)
{
//throw new OoxValidatorException("The file \"" + target + "\" is described in the \"/word/_rels/document.xml.rels\" file but does not exist in the package.");
fileExists = false;
}
if (fileExists)
{
// 5.1. A content type can be found in [Content_Types].xml file
String ct = this.findContentType(reader, "/" + target);
// 5.2. If it's an xml file, it has to be valid
if (ct.EndsWith("+xml"))
{
this.validateXml(item);
}
}
}
}
}
// 6. do all referenced relationships exist in the part relationship file
// retrieve all ids referenced in the document
Stream doc = reader.GetEntry(docTarget);
r = XmlReader.Create(doc);
ArrayList ids = new ArrayList();
while (r.Read())
{
if (r.NodeType == XmlNodeType.Element && r.GetAttribute("id", OOX_DOC_REL_NS) != null)
{
ids.Add(r.GetAttribute("id", OOX_DOC_REL_NS));
}
}
// check if each id exists in the partRel file
if (ids.Count != 0)
{
if (!partRelExists)
{
throw new ValidationException("Referenced id exist but no part relationship file found");
}
relationShips = reader.GetEntry(partRelPath);
r = XmlReader.Create(relationShips);
while (r.Read())
{
if (r.NodeType == XmlNodeType.Element && r.LocalName == "Relationship")
{
if (ids.Contains(r.GetAttribute("Id")))
{
ids.Remove(r.GetAttribute("Id"));
}
}
}
if (ids.Count != 0)
{
throw new ValidationException("One or more relationship id have not been found in the partRelationship file : " + ids[0]);
}
}
}