private static object RemoveCustomXmlAndContentControlsTransform(
XNode node, SimplifyMarkupSettings simplifyMarkupSettings)
{
XElement element = node as XElement;
if (element != null)
{
if (simplifyMarkupSettings.RemoveSmartTags &&
element.Name == W.smartTag)
{
return(element
.Elements()
.Select(e =>
RemoveCustomXmlAndContentControlsTransform(e,
simplifyMarkupSettings)));
}
if (simplifyMarkupSettings.RemoveContentControls &&
element.Name == W.sdt)
{
return(element
.Element(W.sdtContent)
.Elements()
.Select(e =>
RemoveCustomXmlAndContentControlsTransform(e,
simplifyMarkupSettings)));
}
}
return(node);
}
public static WmlDocument SimplifyMarkup(WmlDocument doc, SimplifyMarkupSettings settings)
{
using (OpenXmlMemoryStreamDocument streamDoc = new OpenXmlMemoryStreamDocument(doc))
{
using (WordprocessingDocument document = streamDoc.GetWordprocessingDocument())
{
SimplifyMarkup(document, settings);
}
return streamDoc.GetModifiedWmlDocument();
}
}
public static WmlDocument SimplifyMarkup(WmlDocument doc, SimplifyMarkupSettings settings)
{
using (OpenXmlMemoryStreamDocument streamDoc = new OpenXmlMemoryStreamDocument(doc))
{
using (WordprocessingDocument document = streamDoc.GetWordprocessingDocument())
{
SimplifyMarkup(document, settings);
}
return(streamDoc.GetModifiedWmlDocument());
}
}
public static void SimplifyMarkup(WordprocessingDocument doc,
SimplifyMarkupSettings settings)
{
if (settings.AcceptRevisions)
RevisionAccepter.AcceptRevisions(doc);
foreach (var part in doc.ContentParts())
SimplifyMarkupForPart(part, settings);
if (doc.MainDocumentPart.StyleDefinitionsPart != null)
SimplifyMarkupForPart(doc.MainDocumentPart.StyleDefinitionsPart, settings);
if (doc.MainDocumentPart.StylesWithEffectsPart != null)
SimplifyMarkupForPart(doc.MainDocumentPart.StylesWithEffectsPart, settings);
}
public static void SimplifyMarkup(WordprocessingDocument doc,
SimplifyMarkupSettings settings)
{
if (settings.AcceptRevisions)
{
RevisionAccepter.AcceptRevisions(doc);
}
foreach (var part in doc.ContentParts())
{
SimplifyMarkupForPart(part, settings);
}
if (doc.MainDocumentPart.StyleDefinitionsPart != null)
{
SimplifyMarkupForPart(doc.MainDocumentPart.StyleDefinitionsPart, settings);
}
if (doc.MainDocumentPart.StylesWithEffectsPart != null)
{
SimplifyMarkupForPart(doc.MainDocumentPart.StylesWithEffectsPart, settings);
}
}
public WmlDocument SimplifyMarkup(SimplifyMarkupSettings settings)
{
return MarkupSimplifier.SimplifyMarkup(this, settings);
}
// lastRenderedPageBreak, permEnd, permStart, proofErr, noProof
// softHyphen:
// Remove when simplifying.
// fldSimple, fldData, fldChar, instrText:
// For hyperlinks, generate same in XHtml. Other than hyperlinks, do the following:
// - collapse fldSimple
// - remove fldSimple, fldData, fldChar, instrText.
private static object SimplifyMarkupTransform(
XNode node,
SimplifyMarkupSettings settings,
SimplifyMarkupParameters parameters)
{
XElement element = node as XElement;
if (element != null)
{
if (settings.RemovePermissions &&
(element.Name == W.permEnd ||
element.Name == W.permStart))
return null;
if (settings.RemoveProof &&
(element.Name == W.proofErr ||
element.Name == W.noProof))
return null;
if (settings.RemoveSoftHyphens &&
element.Name == W.softHyphen)
return null;
if (settings.RemoveLastRenderedPageBreak &&
element.Name == W.lastRenderedPageBreak)
return null;
if (settings.RemoveBookmarks &&
(element.Name == W.bookmarkStart ||
element.Name == W.bookmarkEnd))
return null;
if (settings.RemoveGoBackBookmark &&
((element.Name == W.bookmarkStart && (int)element.Attribute(W.id) == parameters.GoBackId) ||
(element.Name == W.bookmarkEnd && (int)element.Attribute(W.id) == parameters.GoBackId)))
return null;
if (settings.RemoveWebHidden &&
element.Name == W.webHidden)
return null;
if (settings.ReplaceTabsWithSpaces && element.Name == W.tab &&
element.Parent.Name == W.r)
return new XElement(W.t,
new XAttribute(XNamespace.Xml + "space", "preserve"),
" ");
if (settings.RemoveComments &&
(element.Name == W.commentRangeStart ||
element.Name == W.commentRangeEnd ||
element.Name == W.commentReference ||
element.Name == W.annotationRef))
return null;
if (settings.RemoveComments &&
element.Name == W.rStyle &&
element.Attribute(W.val).Value == "CommentReference")
return null;
if (settings.RemoveEndAndFootNotes &&
(element.Name == W.endnoteReference ||
element.Name == W.footnoteReference))
return null;
if (settings.RemoveFieldCodes)
{
if (element.Name == W.fldSimple)
return element.Elements().Select(e =>
SimplifyMarkupTransform(e, settings, parameters));
if (element.Name == W.fldData ||
element.Name == W.fldChar ||
element.Name == W.instrText)
return null;
}
return new XElement(element.Name,
element.Attributes(),
element.Nodes().Select(n =>
SimplifyMarkupTransform(n, settings, parameters)));
}
return node;
}
private static void SimplifyMarkupForPart(
OpenXmlPart part,
SimplifyMarkupSettings settings)
{
SimplifyMarkupParameters parameters = new SimplifyMarkupParameters();
if (part.ContentType == "application/vnd.openxmlformats-officedocument.wordprocessingml.document.main+xml")
{
WordprocessingDocument doc = (WordprocessingDocument)part.OpenXmlPackage;
if (settings.RemoveGoBackBookmark == true)
{
var goBackBookmark = doc
.MainDocumentPart
.GetXDocument()
.Root
.Descendants(W.bookmarkStart)
.FirstOrDefault(bm => (string)bm.Attribute(W.name) == "_GoBack");
if (goBackBookmark != null)
parameters.GoBackId = (int)goBackBookmark.Attribute(W.id);
}
}
XDocument xdoc = part.GetXDocument();
XElement newRoot = xdoc.Root;
// Need to do this first to enable simplifying hyperlinks.
if (settings.RemoveContentControls ||
settings.RemoveSmartTags)
newRoot = (XElement)
RemoveCustomXmlAndContentControlsTransform(
newRoot, settings);
// This may touch many elements, so needs to be its own
// transform.
if (settings.RemoveRsidInfo)
newRoot = (XElement)RemoveRsidTransform(newRoot);
XDocument prevNewRoot = new XDocument(newRoot);
while (true)
{
if (settings.RemoveComments ||
settings.RemoveEndAndFootNotes ||
settings.ReplaceTabsWithSpaces ||
settings.RemoveFieldCodes ||
settings.RemovePermissions ||
settings.RemoveProof ||
settings.RemoveBookmarks ||
settings.RemoveWebHidden ||
settings.RemoveGoBackBookmark)
newRoot = (XElement)SimplifyMarkupTransform(newRoot,
settings, parameters);
// Remove runs and run properties that have become empty due to previous
// transforms.
newRoot = (XElement)
RemoveEmptyRunsAndRunPropertiesTransform(newRoot);
// Merge adjacent runs that have identical run properties.
newRoot = (XElement)MergeAdjacentRunsTransform(newRoot);
// Merge adjacent instrText elements.
newRoot = (XElement)MergeAdjacentInstrText(newRoot);
if (XNode.DeepEquals(prevNewRoot.Root, newRoot))
break;
prevNewRoot = new XDocument(newRoot);
}
if (settings.NormalizeXml)
{
XAttribute[] ns_attrs =
{
new XAttribute(XNamespace.Xmlns + "wpc", WPC.wpc),
new XAttribute(XNamespace.Xmlns + "mc", MC.mc),
new XAttribute(XNamespace.Xmlns + "o", O.o),
new XAttribute(XNamespace.Xmlns + "r", R.r),
new XAttribute(XNamespace.Xmlns + "m", M.m),
new XAttribute(XNamespace.Xmlns + "v", VML.vml),
new XAttribute(XNamespace.Xmlns + "wp14", WP14.wp14),
new XAttribute(XNamespace.Xmlns + "wp", WP.wp),
new XAttribute(XNamespace.Xmlns + "w10", W10.w10),
new XAttribute(XNamespace.Xmlns + "w", W.w),
new XAttribute(XNamespace.Xmlns + "w14", W14.w14),
new XAttribute(XNamespace.Xmlns + "wpg", WPG.wpg),
new XAttribute(XNamespace.Xmlns + "wpi", WPI.wpi),
new XAttribute(XNamespace.Xmlns + "wne", WNE.wne),
new XAttribute(XNamespace.Xmlns + "wps", WPS.wps),
new XAttribute(MC.Ignorable, "w14 wp14"),
};
XDocument newXDoc = Normalize(new XDocument(newRoot), null);
foreach (var nsatt in ns_attrs)
{
if (newXDoc.Root.Attribute(nsatt.Name) == null)
newXDoc.Root.Add(nsatt);
}
part.PutXDocument(newXDoc);
}
else
{
part.PutXDocument(new XDocument(newRoot));
}
}
private static object RemoveCustomXmlAndContentControlsTransform(
XNode node, SimplifyMarkupSettings simplifyMarkupSettings)
{
XElement element = node as XElement;
if (element != null)
{
if (simplifyMarkupSettings.RemoveSmartTags &&
element.Name == W.smartTag)
return element
.Elements()
.Select(e =>
RemoveCustomXmlAndContentControlsTransform(e,
simplifyMarkupSettings));
if (simplifyMarkupSettings.RemoveContentControls &&
element.Name == W.sdt)
return element
.Element(W.sdtContent)
.Elements()
.Select(e =>
RemoveCustomXmlAndContentControlsTransform(e,
simplifyMarkupSettings));
}
return node;
}
private static void SimplifyMarkupForPart(
OpenXmlPart part,
SimplifyMarkupSettings settings)
{
SimplifyMarkupParameters parameters = new SimplifyMarkupParameters();
if (part.ContentType == "application/vnd.openxmlformats-officedocument.wordprocessingml.document.main+xml")
{
WordprocessingDocument doc = (WordprocessingDocument)part.OpenXmlPackage;
if (settings.RemoveGoBackBookmark == true)
{
var goBackBookmark = doc
.MainDocumentPart
.GetXDocument()
.Root
.Descendants(W.bookmarkStart)
.FirstOrDefault(bm => (string)bm.Attribute(W.name) == "_GoBack");
if (goBackBookmark != null)
{
parameters.GoBackId = (int)goBackBookmark.Attribute(W.id);
}
}
}
XDocument xdoc = part.GetXDocument();
XElement newRoot = xdoc.Root;
// Need to do this first to enable simplifying hyperlinks.
if (settings.RemoveContentControls ||
settings.RemoveSmartTags)
{
newRoot = (XElement)
RemoveCustomXmlAndContentControlsTransform(
newRoot, settings);
}
// This may touch many elements, so needs to be its own
// transform.
if (settings.RemoveRsidInfo)
{
newRoot = (XElement)RemoveRsidTransform(newRoot);
}
XDocument prevNewRoot = new XDocument(newRoot);
while (true)
{
if (settings.RemoveComments ||
settings.RemoveEndAndFootNotes ||
settings.ReplaceTabsWithSpaces ||
settings.RemoveFieldCodes ||
settings.RemovePermissions ||
settings.RemoveProof ||
settings.RemoveBookmarks ||
settings.RemoveWebHidden ||
settings.RemoveGoBackBookmark)
{
newRoot = (XElement)SimplifyMarkupTransform(newRoot,
settings, parameters);
}
// Remove runs and run properties that have become empty due to previous
// transforms.
newRoot = (XElement)
RemoveEmptyRunsAndRunPropertiesTransform(newRoot);
// Merge adjacent runs that have identical run properties.
newRoot = (XElement)MergeAdjacentRunsTransform(newRoot);
// Merge adjacent instrText elements.
newRoot = (XElement)MergeAdjacentInstrText(newRoot);
if (XNode.DeepEquals(prevNewRoot.Root, newRoot))
{
break;
}
prevNewRoot = new XDocument(newRoot);
}
if (settings.NormalizeXml)
{
XAttribute[] ns_attrs =
{
new XAttribute(XNamespace.Xmlns + "wpc", WPC.wpc),
new XAttribute(XNamespace.Xmlns + "mc", MC.mc),
new XAttribute(XNamespace.Xmlns + "o", O.o),
new XAttribute(XNamespace.Xmlns + "r", R.r),
new XAttribute(XNamespace.Xmlns + "m", M.m),
new XAttribute(XNamespace.Xmlns + "v", VML.vml),
new XAttribute(XNamespace.Xmlns + "wp14", WP14.wp14),
new XAttribute(XNamespace.Xmlns + "wp", WP.wp),
new XAttribute(XNamespace.Xmlns + "w10", W10.w10),
new XAttribute(XNamespace.Xmlns + "w", W.w),
new XAttribute(XNamespace.Xmlns + "w14", W14.w14),
new XAttribute(XNamespace.Xmlns + "wpg", WPG.wpg),
new XAttribute(XNamespace.Xmlns + "wpi", WPI.wpi),
new XAttribute(XNamespace.Xmlns + "wne", WNE.wne),
new XAttribute(XNamespace.Xmlns + "wps", WPS.wps),
new XAttribute(MC.Ignorable, "w14 wp14"),
};
XDocument newXDoc = Normalize(new XDocument(newRoot), null);
foreach (var nsatt in ns_attrs)
{
if (newXDoc.Root.Attribute(nsatt.Name) == null)
{
newXDoc.Root.Add(nsatt);
}
}
part.PutXDocument(newXDoc);
}
else
{
part.PutXDocument(new XDocument(newRoot));
}
}
// lastRenderedPageBreak, permEnd, permStart, proofErr, noProof
// softHyphen:
// Remove when simplifying.
// fldSimple, fldData, fldChar, instrText:
// For hyperlinks, generate same in XHtml. Other than hyperlinks, do the following:
// - collapse fldSimple
// - remove fldSimple, fldData, fldChar, instrText.
private static object SimplifyMarkupTransform(
XNode node,
SimplifyMarkupSettings settings,
SimplifyMarkupParameters parameters)
{
XElement element = node as XElement;
if (element != null)
{
if (settings.RemovePermissions &&
(element.Name == W.permEnd ||
element.Name == W.permStart))
{
return(null);
}
if (settings.RemoveProof &&
(element.Name == W.proofErr ||
element.Name == W.noProof))
{
return(null);
}
if (settings.RemoveSoftHyphens &&
element.Name == W.softHyphen)
{
return(null);
}
if (settings.RemoveLastRenderedPageBreak &&
element.Name == W.lastRenderedPageBreak)
{
return(null);
}
if (settings.RemoveBookmarks &&
(element.Name == W.bookmarkStart ||
element.Name == W.bookmarkEnd))
{
return(null);
}
if (settings.RemoveGoBackBookmark &&
((element.Name == W.bookmarkStart && (int)element.Attribute(W.id) == parameters.GoBackId) ||
(element.Name == W.bookmarkEnd && (int)element.Attribute(W.id) == parameters.GoBackId)))
{
return(null);
}
if (settings.RemoveWebHidden &&
element.Name == W.webHidden)
{
return(null);
}
if (settings.ReplaceTabsWithSpaces && element.Name == W.tab &&
element.Parent.Name == W.r)
{
return(new XElement(W.t,
new XAttribute(XNamespace.Xml + "space", "preserve"),
" "));
}
if (settings.RemoveComments &&
(element.Name == W.commentRangeStart ||
element.Name == W.commentRangeEnd ||
element.Name == W.commentReference ||
element.Name == W.annotationRef))
{
return(null);
}
if (settings.RemoveComments &&
element.Name == W.rStyle &&
element.Attribute(W.val).Value == "CommentReference")
{
return(null);
}
if (settings.RemoveEndAndFootNotes &&
(element.Name == W.endnoteReference ||
element.Name == W.footnoteReference))
{
return(null);
}
if (settings.RemoveFieldCodes)
{
if (element.Name == W.fldSimple)
{
return(element.Elements().Select(e =>
SimplifyMarkupTransform(e, settings, parameters)));
}
if (element.Name == W.fldData ||
element.Name == W.fldChar ||
element.Name == W.instrText)
{
return(null);
}
}
return(new XElement(element.Name,
element.Attributes(),
element.Nodes().Select(n =>
SimplifyMarkupTransform(n, settings, parameters))));
}
return(node);
}
public WmlDocument SimplifyMarkup(SimplifyMarkupSettings settings)
{
return(MarkupSimplifier.SimplifyMarkup(this, settings));
}