private void process(IElement node, bool root)
{
IList<RDFa.IncompleteTriple> incompleteTriplesLocal=new List<RDFa.IncompleteTriple>();
string localLanguage=context.language;
RDFTerm newSubject=null;
bool recurse=true;
bool skipElement=false;
RDFTerm currentObject=null;
IDictionary<string,string> namespacesLocal=
new PeterO.Support.LenientDictionary<string,string>(context.namespaces);
IDictionary<string,string> iriMapLocal=
new PeterO.Support.LenientDictionary<string,string>(context.iriMap);
string attr=null;
if(!xhtml){
attr=node.getAttribute("xml:base");
if(attr!=null){
context.baseURI=URIUtility.relativeResolve(attr, context.baseURI);
}
}
// Support XML namespaces
foreach(var attrib in node.getAttributes()){
string name=StringUtility.toLowerCaseAscii(attrib.getName());
//Console.WriteLine(attrib);
if(name.Equals("xmlns")){
//Console.WriteLine("xmlns %s",attrib.getValue());
iriMapLocal.Add("", attrib.getValue());
namespacesLocal.Add("", attrib.getValue());
} else if(name.StartsWith("xmlns:",StringComparison.Ordinal) && name.Length>6){
string prefix=name.Substring(6);
//Console.WriteLine("xmlns %s %s",prefix,attrib.getValue());
if(!"_".Equals(prefix)){
iriMapLocal.Add(prefix, attrib.getValue());
}
namespacesLocal.Add(prefix, attrib.getValue());
}
}
attr=node.getAttribute("xml:lang");
if(attr!=null){
localLanguage=attr;
}
// Support RDF/XML metadata
if(node.getLocalName().Equals("RDF") &&
RDF_NAMESPACE.Equals(node.getNamespaceURI())){
miniRdfXml(node,context);
return;
}
string rel=node.getAttribute("rel");
string rev=node.getAttribute("rev");
string property=node.getAttribute("property");
string content=node.getAttribute("content");
string datatype=node.getAttribute("datatype");
if(rel==null && rev==null){
// Step 4
RDFTerm resource=getSafeCurieOrCurieOrIri(
node.getAttribute("about"),iriMapLocal);
if(resource==null){
resource=getSafeCurieOrCurieOrIri(
node.getAttribute("resource"),iriMapLocal);
}
if(resource==null){
resource=relativeResolve(node.getAttribute("href"));
}
if(resource==null){
resource=relativeResolve(node.getAttribute("src"));
}
if((resource==null || resource.getKind()!=RDFTerm.IRI)){
string rdfTypeof=getCurie(node.getAttribute("typeof"),iriMapLocal);
if(isHtmlElement(node, "head") ||
isHtmlElement(node, "body")){
resource=getSafeCurieOrCurieOrIri("",iriMapLocal);
}
if(resource==null && !xhtml && root){
resource=getSafeCurieOrCurieOrIri("",iriMapLocal);
}
if(resource==null && rdfTypeof!=null){
resource=generateBlankNode();
}
if(resource==null){
if(context.parentObject!=null) {
resource=context.parentObject;
}
if(node.getAttribute("property")==null){
skipElement=true;
}
}
newSubject=resource;
} else {
newSubject=resource;
}
} else {
// Step 5
RDFTerm resource=getSafeCurieOrCurieOrIri(
node.getAttribute("about"),iriMapLocal);
if(resource==null){
resource=relativeResolve(node.getAttribute("src"));
}
if((resource==null || resource.getKind()!=RDFTerm.IRI)){
string rdfTypeof=getCurie(node.getAttribute("typeof"),iriMapLocal);
if(isHtmlElement(node, "head") ||
isHtmlElement(node, "body")){
resource=getSafeCurieOrCurieOrIri("",iriMapLocal);
}
if(resource==null && !xhtml && root){
resource=getSafeCurieOrCurieOrIri("",iriMapLocal);
}
if(resource==null && rdfTypeof!=null){
resource=generateBlankNode();
}
if(resource==null){
if(context.parentObject!=null) {
resource=context.parentObject;
}
}
newSubject=resource;
} else {
newSubject=resource;
}
resource=getSafeCurieOrCurieOrIri(
node.getAttribute("resource"),iriMapLocal);
if(resource==null){
resource=relativeResolve(node.getAttribute("href"));
}
currentObject=resource;
}
// Step 6
if(newSubject!=null){
string[] types=StringUtility.splitAtNonFFSpaces(node.getAttribute("typeof"));
foreach(var type in types){
string iri=getCurie(type,iriMapLocal);
if(iri!=null){
outputGraph.Add(new RDFTriple(
newSubject,RDFTerm.A,
RDFTerm.fromIRI(iri)
));
}
}
}
// Step 7
if(currentObject!=null){
string[] types=StringUtility.splitAtNonFFSpaces(rel);
foreach(var type in types){
string iri=getRelTermOrCurie(type,
iriMapLocal);
#if DEBUG
if(!(newSubject!=null))throw new InvalidOperationException("doesn't satisfy newSubject!=null");
#endif
if(iri!=null){
outputGraph.Add(new RDFTriple(
newSubject,
RDFTerm.fromIRI(iri),currentObject
));
}
}
types=StringUtility.splitAtNonFFSpaces(rev);
foreach(var type in types){
string iri=getRelTermOrCurie(type,
iriMapLocal);
if(iri!=null){
outputGraph.Add(new RDFTriple(
currentObject,
RDFTerm.fromIRI(iri),
newSubject
));
}
}
} else {
// Step 8
string[] types=StringUtility.splitAtNonFFSpaces(rel);
bool hasPredicates=false;
// Defines predicates
foreach(var type in types){
string iri=getRelTermOrCurie(type,
iriMapLocal);
if(iri!=null){
if(!hasPredicates){
hasPredicates=true;
currentObject=generateBlankNode();
}
RDFa.IncompleteTriple inc=new RDFa.IncompleteTriple();
inc.predicate=RDFTerm.fromIRI(iri);
inc.direction=RDFa.ChainingDirection.Forward;
incompleteTriplesLocal.Add(inc);
}
}
types=StringUtility.splitAtNonFFSpaces(rev);
foreach(var type in types){
string iri=getRelTermOrCurie(type,
iriMapLocal);
if(iri!=null){
if(!hasPredicates){
hasPredicates=true;
currentObject=generateBlankNode();
}
RDFa.IncompleteTriple inc=new RDFa.IncompleteTriple();
inc.predicate=RDFTerm.fromIRI(iri);
inc.direction=RDFa.ChainingDirection.Reverse;
incompleteTriplesLocal.Add(inc);
}
}
}
// Step 9
string[] preds=StringUtility.splitAtNonFFSpaces(property);
string datatypeValue=getCurie(datatype,
iriMapLocal);
if(datatype!=null && datatypeValue==null) {
datatypeValue="";
}
//Console.WriteLine("datatype=[%s] prop=%s vocab=%s",
// datatype,property,localDefaultVocab);
//Console.WriteLine("datatypeValue=[%s]",datatypeValue);
RDFTerm currentProperty=null;
foreach(var pred in preds){
string iri=getCurie(pred,
iriMapLocal);
if(iri!=null){
//Console.WriteLine("iri=[%s]",iri);
currentProperty=null;
if(datatypeValue!=null && datatypeValue.Length>0 &&
!datatypeValue.Equals(RDF_XMLLITERAL)){
string literal=content;
if(literal==null) {
literal=getTextNodeText(node);
}
currentProperty=RDFTerm.fromTypedString(literal,datatypeValue);
} else if(node.getAttribute("content")!=null ||
!hasNonTextChildNodes(node) ||
(datatypeValue!=null && datatypeValue.Length==0)){
string literal=node.getAttribute("content");
if(literal==null) {
literal=getTextNodeText(node);
}
currentProperty=(!string.IsNullOrEmpty(localLanguage)) ?
RDFTerm.fromLangString(literal, localLanguage) :
RDFTerm.fromTypedString(literal);
} else if(hasNonTextChildNodes(node) &&
(datatypeValue==null || datatypeValue.Equals(RDF_XMLLITERAL))){
// XML literal
recurse=false;
if(datatypeValue==null) {
datatypeValue=RDF_XMLLITERAL;
}
try {
string literal=ExclusiveCanonicalXML.canonicalize(node,
false, namespacesLocal);
currentProperty=RDFTerm.fromTypedString(literal,datatypeValue);
} catch(ArgumentException){
// failure to canonicalize
}
}
#if DEBUG
if(!(newSubject!=null))throw new InvalidOperationException("doesn't satisfy newSubject!=null");
#endif
outputGraph.Add(new RDFTriple(
newSubject,
RDFTerm.fromIRI(iri),currentProperty
));
}
}
// Step 10
if(!skipElement && newSubject!=null){
foreach(var triple in context.incompleteTriples){
if(triple.direction==RDFa.ChainingDirection.Forward){
outputGraph.Add(new RDFTriple(
context.parentSubject,
triple.predicate,
newSubject));
} else {
outputGraph.Add(new RDFTriple(
newSubject,triple.predicate,
context.parentSubject));
}
}
}
// Step 13
if(recurse){
foreach(var childNode in node.getChildNodes()){
IElement childElement;
RDFa.EvalContext oldContext=context;
if(childNode is IElement){
childElement=((IElement)childNode);
//Console.WriteLine("skip=%s vocab=%s local=%s",
// skipElement,context.defaultVocab,
//localDefaultVocab);
if(skipElement){
RDFa.EvalContext ec=oldContext.copy();
ec.language=localLanguage;
ec.iriMap=iriMapLocal;
ec.namespaces=namespacesLocal;
context=ec;
process(childElement,false);
} else {
RDFa.EvalContext ec=new RDFa.EvalContext();
ec.baseURI=oldContext.baseURI;
ec.iriMap=iriMapLocal;
ec.namespaces=namespacesLocal;
ec.incompleteTriples=incompleteTriplesLocal;
ec.parentSubject=((newSubject==null) ? oldContext.parentSubject :
newSubject);
ec.parentObject=((currentObject==null) ?
((newSubject==null) ? oldContext.parentSubject :
newSubject) : currentObject);
ec.language=localLanguage;
context=ec;
process(childElement,false);
}
}
context=oldContext;
}
}
}
// Processes a subset of RDF/XML metadata
// Doesn't implement RDF/XML completely
private void miniRdfXml(IElement node, RDFa.EvalContext context, RDFTerm subject)
{
string language=context.language;
foreach(var child in node.getChildNodes()){
IElement childElement=(child is IElement) ?
((IElement)child) : null;
if(childElement==null) {
continue;
}
if(node.getAttribute("xml:lang")!=null){
language=node.getAttribute("xml:lang");
} else {
language=context.language;
}
if(childElement.getLocalName().Equals("Description") &&
RDF_NAMESPACE.Equals(childElement.getNamespaceURI())){
RDFTerm about=relativeResolve(childElement.getAttributeNS(RDF_NAMESPACE,"about"));
//Console.WriteLine("about=%s [%s]",about,childElement.getAttribute("about"));
if(about==null){
about=subject;
if(about==null) {
continue;
}
}
foreach(var child2 in child.getChildNodes()){
IElement childElement2=
((child2 is IElement) ?
((IElement)child2) : null);
if(childElement2==null) {
continue;
}
miniRdfXmlChild(childElement2,about,language);
}
} else if(RDF_NAMESPACE.Equals(childElement.getNamespaceURI()))
throw new NotSupportedException();
}
}
private void miniRdfXmlChild(IElement node, RDFTerm subject, string language)
{
string nsname=node.getNamespaceURI();
if(node.getAttribute("xml:lang")!=null){
language=node.getAttribute("xml:lang");
}
string localname=node.getLocalName();
RDFTerm predicate=relativeResolve(nsname+localname);
if(!hasNonTextChildNodes(node)){
string content=getTextNodeText(node);
RDFTerm literal;
if(!string.IsNullOrEmpty(language)){
literal=RDFTerm.fromLangString(content, language);
} else {
literal=RDFTerm.fromTypedString(content);
}
outputGraph.Add(new RDFTriple(subject,predicate,literal));
} else {
string parseType=node.getAttributeNS(RDF_NAMESPACE, "parseType");
if("Literal".Equals(parseType))
throw new NotSupportedException();
RDFTerm blank=generateBlankNode();
context.language=language;
miniRdfXml(node,context,blank);
outputGraph.Add(new RDFTriple(subject,predicate,blank));
}
}
private static bool implyForLink(IElement root, JSONObject subProperties)
{
if(StringUtility.toLowerCaseAscii(root.getLocalName()).Equals("a") &&
root.getAttribute("href")!=null){
// get the link's URL
setValueIfAbsent(subProperties,"url", getUValue(root));
IList<IElement> elements=getChildElements(root);
if(elements.Count==1 &&
StringUtility.toLowerCaseAscii(elements[0].getLocalName()).Equals("img")){
string pValue=getPValue(elements[0]); // try to get the ALT/TITLE from the image
if(StringUtility.isNullOrSpaces(pValue))
{
pValue=getPValue(root); // if empty, get text from link instead
}
setValueIfAbsent(subProperties,"name", pValue);
// get the SRC of the image
setValueIfAbsent(subProperties,"photo", getUValue(elements[0]));
} else {
// get the text content
string pvalue=getPValue(root);
if(!StringUtility.isNullOrSpaces(pvalue)) {
setValueIfAbsent(subProperties,"name", pvalue);
}
}
return true;
}
return false;
}
private static void propertyWalk(IElement root,
JSONObject properties, JSONArray children)
{
string[] className=getClassNames(root);
if(className.Length>0){
IList<string> types=new List<string>();
bool hasProperties=false;
foreach(var cls in className){
if(cls.StartsWith("p-",StringComparison.Ordinal) && properties!=null){
hasProperties=true;
} else if(cls.StartsWith("u-",StringComparison.Ordinal) && properties!=null){
hasProperties=true;
} else if(cls.StartsWith("dt-",StringComparison.Ordinal) && properties!=null){
hasProperties=true;
} else if(cls.StartsWith("e-",StringComparison.Ordinal) && properties!=null){
hasProperties=true;
} else if(cls.StartsWith("h-",StringComparison.Ordinal)){
types.Add(cls);
}
}
if(types.Count==0 && hasProperties){
// has properties and isn't a microformat
// root
foreach(var cls in className){
if(cls.StartsWith("p-",StringComparison.Ordinal)){
string value=getPValue(root);
if(!StringUtility.isNullOrSpaces(value)) {
accumulateValue(properties,cls.Substring(2),value);
}
} else if(cls.StartsWith("u-",StringComparison.Ordinal)){
accumulateValue(properties,cls.Substring(2),
getUValue(root));
} else if(cls.StartsWith("dt-",StringComparison.Ordinal)){
accumulateValue(properties,cls.Substring(3),
getDTValue(root,getLastKnownTime(properties)));
} else if(cls.StartsWith("e-",StringComparison.Ordinal)){
accumulateValue(properties,cls.Substring(2),
getEValue(root));
}
}
} else if(types.Count>0){
// this is a child microformat
// with no properties
JSONObject obj=new JSONObject();
obj.put("type", new JSONArray(types));
// for holding child elements with
// properties
JSONObject subProperties=new JSONObject();
// for holding child microformats with no
// property class
JSONArray subChildren=new JSONArray();
foreach(var child in root.getChildNodes()){
if(child is IElement) {
propertyWalk((IElement)child,
subProperties,subChildren);
}
}
if(subChildren.Length>0){
obj.put("children", subChildren);
}
if(types.Count>0){
// we imply missing properties here
// Imply p-name and p-url
if(!implyForLink(root,subProperties)){
if(hasSingleChildElementNamed(root,"a")){
implyForLink(getFirstChildElement(root),subProperties);
} else {
string pvalue=getPValue(root);
if(!StringUtility.isNullOrSpaces(pvalue)) {
setValueIfAbsent(subProperties,"name", pvalue);
}
}
}
// Also imply u-photo
if(StringUtility.toLowerCaseAscii(root.getLocalName()).Equals("img") &&
root.getAttribute("src")!=null){
setValueIfAbsent(subProperties,"photo", getUValue(root));
}
if(!subProperties.has("photo")){
IList<IElement> images=root.getElementsByTagName("img");
// If there is only one descendant image, imply
// u-photo
if(images.Count==1){
setValueIfAbsent(subProperties,"photo",
getUValue(images[0]));
}
}
}
obj.put("properties", subProperties);
if(hasProperties){
foreach(var cls in className){
if(cls.StartsWith("p-",StringComparison.Ordinal)){ // property
JSONObject clone=copyJson(obj);
clone.put("value",getPValue(root));
accumulateValue(properties,cls.Substring(2),clone);
} else if(cls.StartsWith("u-",StringComparison.Ordinal)){ // URL
JSONObject clone=copyJson(obj);
clone.put("value",getUValue(root));
accumulateValue(properties,cls.Substring(2),clone);
} else if(cls.StartsWith("dt-",StringComparison.Ordinal)){ // date/time
JSONObject clone=copyJson(obj);
clone.put("value",getDTValue(root,getLastKnownTime(properties)));
accumulateValue(properties,cls.Substring(3),clone);
} else if(cls.StartsWith("e-",StringComparison.Ordinal)){ // date/time
JSONObject clone=copyJson(obj);
clone.put("value",getEValue(root));
accumulateValue(properties,cls.Substring(2),clone);
}
}
} else {
children.put(obj);
}
return;
}
}
foreach(var child in root.getChildNodes()){
if(child is IElement) {
propertyWalk((IElement)child,properties,children);
}
}
}
private static string getValueElementContent(IElement valueElement)
{
if(hasClassName(valueElement,"value-title"))
// If element has the value-title class, use
// the title instead
return valueOrEmpty(valueElement.getAttribute("title"));
else if(elementName(valueElement).Equals("img") ||
elementName(valueElement).Equals("area")){
string s=valueElement.getAttribute("alt");
return (s==null) ? "" : s;
} else if(elementName(valueElement).Equals("data")){
string s=valueElement.getAttribute("value");
return (s==null) ? getTrimmedTextContent(valueElement) : s;
} else if(elementName(valueElement).Equals("abbr")){
string s=valueElement.getAttribute("title");
return (s==null) ? getTrimmedTextContent(valueElement) : s;
} else
return getTrimmedTextContent(valueElement);
}
private static bool hasClassName(IElement e, string className)
{
string attr=e.getAttribute("class");
if(attr==null || attr.Length<className.Length)return false;
string[] cls=StringUtility.splitAtSpaces(attr);
foreach(var c in cls){
if(c.Equals(className))return true;
}
return false;
}
private static string[] getRelNames(IElement element)
{
string[] ret=StringUtility.splitAtSpaces(
StringUtility.toLowerCaseAscii(element.getAttribute("rel")));
if(ret.Length==0)return ret;
IList<string> retList=new List<string>();
foreach(var element2 in ret) {
retList.Add(element2);
}
if(retList.Count>=2){
ISet<string> stringSet=new HashSet<string>(retList);
return PeterO.Support.Collections.ToArray(stringSet);
} else
return retList.ToArray();
}
private static string getPValue(IElement root)
{
if(root.getAttribute("title")!=null)
return root.getAttribute("title");
if(StringUtility.toLowerCaseAscii(root.getLocalName()).Equals("img") &&
!StringUtility.isNullOrSpaces(root.getAttribute("alt")))
return root.getAttribute("alt");
return getValueContent(root,false);
}
private static string getHref(IElement node)
{
string name=StringUtility.toLowerCaseAscii(node.getLocalName());
string href="";
if("a".Equals(name) || "link".Equals(name) || "area".Equals(name)){
href=node.getAttribute("href");
} else if("object".Equals(name)){
href=node.getAttribute("data");
} else if("img".Equals(name) || "source".Equals(name) ||
"track".Equals(name) ||
"iframe".Equals(name) ||
"audio".Equals(name) ||
"video".Equals(name) ||
"embed".Equals(name)){
href=node.getAttribute("src");
} else
return null;
if(href==null || href.Length==0)
return "";
href=HtmlDocument.resolveURL(node,href,null);
if(href==null || href.Length==0)
return "";
return href;
}
private static string getDTValueContent(IElement valueElement)
{
string elname=elementName(valueElement);
string text="";
if(hasClassName(valueElement,"value-title"))
return valueOrEmpty(valueElement.getAttribute("title"));
else if(elname.Equals("img") || elname.Equals("area")){
string s=valueElement.getAttribute("alt");
text=(s==null) ? "" : s;
} else if(elname.Equals("data")){
string s=valueElement.getAttribute("value");
text=(s==null) ? getTrimmedTextContent(valueElement) : s;
} else if(elname.Equals("abbr")){
string s=valueElement.getAttribute("title");
text=(s==null) ? getTrimmedTextContent(valueElement) : s;
} else if(elname.Equals("del") || elname.Equals("ins") || elname.Equals("time")){
string s=valueElement.getAttribute("datetime");
if(StringUtility.isNullOrSpaces(s)) {
s=valueElement.getAttribute("title");
}
text=(s==null) ? getTrimmedTextContent(valueElement) : s;
} else {
text=getTrimmedTextContent(valueElement);
}
return text;
}
private static string[] getClassNames(IElement element)
{
string[] ret=StringUtility.splitAtSpaces(element.getAttribute("class"));
string[] rel=parseLegacyRel(element.getAttribute("rel"));
if(ret.Length==0 && rel.Length==0)return ret;
// Replace old microformats class names with
// their modern versions
IList<string> retList=new List<string>();
foreach(var element2 in rel) {
retList.Add(element2);
}
foreach(var element2 in ret) {
string legacyLabel=legacyLabelsMap[element2];
if(complexLegacyMap.ContainsKey(element2)){
foreach(var item in complexLegacyMap[element2]){
retList.Add(item);
}
}
else if(legacyLabel!=null) {
retList.Add(legacyLabel);
} else {
retList.Add(element2);
}
}
if(retList.Count>=2){
ISet<string> stringSet=new HashSet<string>(retList);
return PeterO.Support.Collections.ToArray(stringSet);
} else
return retList.ToArray();
}
