added support for XML processing instructions

htmlparser/src/CNavDTD.cpp

@@ -199,6 +199,7 @@ CToken* nsCTokenRecycler::CreateTokenOfType(eHTMLTokenTypes aType,eHTMLTags aTag
       case eToken_script:     result=new CScriptToken(); break;
       case eToken_style:      result=new CStyleToken(); break;
       case eToken_skippedcontent: result=new CSkippedContentToken(aString); break;
+      case eToken_instruction:result=new CInstructionToken(); break;
         default:
           break;
     }
@@ -308,6 +309,8 @@ PRInt32 NavDispatchTokenHandler(CToken* aToken,nsIDTD* aDTD) {
         result=theDTD->HandleStyleToken(aToken); break;
       case eToken_skippedcontent:
         result=theDTD->HandleSkippedContentToken(aToken); break;
+      case eToken_instruction:
+        result=theDTD->HandleProcessingInstructionToken(aToken); break;
       default:
         result=0;
     }//switch
@@ -337,6 +340,7 @@ void CNavDTD::InitializeDefaultTokenHandlers() {
 //  AddTokenHandler(new CTokenHandler(NavDispatchTokenHandler,eToken_script));
   AddTokenHandler(new CTokenHandler(NavDispatchTokenHandler,eToken_style));
   AddTokenHandler(new CTokenHandler(NavDispatchTokenHandler,eToken_skippedcontent));
+  AddTokenHandler(new CTokenHandler(NavDispatchTokenHandler,eToken_instruction));
 }
 
 
@@ -864,6 +868,21 @@ nsresult CNavDTD::HandleStyleToken(CToken* aToken){
 }
 
 
+/**
+ *  This method gets called when an "instruction" token has been 
+ *  encountered in the parse process. 
+ *  
+ *  @update  gess 3/25/98
+ *  @param   aToken -- next (start) token to be handled
+ *  @return  PR_TRUE if all went well; PR_FALSE if error occured
+ */
+nsresult CNavDTD::HandleProcessingInstructionToken(CToken* aToken){
+  NS_PRECONDITION(0!=aToken,kNullToken);
+
+//  CStyleToken*  st = (CStyleToken*)(aToken);
+  return NS_OK;
+}
+
 /**
  * Retrieve the attributes for this node, and add then into
  * the node.
@@ -2698,9 +2717,15 @@ CNavDTD::ConsumeTag(PRUnichar aChar,CScanner& aScanner,CToken*& aToken) {
           else aToken=gTokenRecycler.CreateTokenOfType(eToken_comment,eHTMLTag_unknown,gEmpty);
         }//if
         break;
+
       case kExclamation:
         aToken=gTokenRecycler.CreateTokenOfType(eToken_comment,eHTMLTag_comment,gEmpty);
         break;
+
+      case kQuestionMark: //it must be an XML processing instruction...
+        aToken=gTokenRecycler.CreateTokenOfType(eToken_instruction,eHTMLTag_unknown,gEmpty);
+        break;
+
       default:
         if(nsString::IsAlpha(aChar))
           return ConsumeStartTag(aChar,aScanner,aToken);

htmlparser/src/CNavDTD.h

@@ -416,6 +416,7 @@ CLASS_EXPORT_HTMLPARS CNavDTD : public nsIDTD {
     nsresult HandleAttributeToken(CToken* aToken);
     nsresult HandleScriptToken(nsCParserNode& aNode);
     nsresult HandleStyleToken(CToken* aToken);
+    nsresult HandleProcessingInstructionToken(CToken* aToken);
 
     virtual  nsITokenRecycler* GetTokenRecycler(void);
 

htmlparser/src/nsHTMLTokens.cpp

@@ -1290,3 +1290,60 @@ const char* GetTagName(PRInt32 aTag) {
   }
   return result;
 }
+
+
+/**
+ *  
+ *  
+ *  @update  gess 9/23/98
+ *  @param   
+ *  @return  
+ */
+CInstructionToken::CInstructionToken() : CHTMLToken(eHTMLTag_unknown) {
+}
+
+/**
+ *  
+ *  
+ *  @update  gess 9/23/98
+ *  @param   
+ *  @return  
+ */
+CInstructionToken::CInstructionToken(const nsString& aString) : CHTMLToken(aString) {
+}
+
+/**
+ *  
+ *  
+ *  @update  gess 9/23/98
+ *  @param   
+ *  @return  
+ */
+nsresult CInstructionToken::Consume(PRUnichar aChar,CScanner& aScanner){
+  mTextValue="<?";
+  nsresult result=aScanner.ReadUntil(mTextValue,kGreaterThan,PR_TRUE);
+  return result;
+}
+
+/**
+ *  
+ *  
+ *  @update  gess 9/23/98
+ *  @param   
+ *  @return  
+ */
+const char* CInstructionToken::GetClassName(void){
+  return "instruction";
+}
+
+/**
+ *  
+ *  
+ *  @update  gess 9/23/98
+ *  @param   
+ *  @return  
+ */
+PRInt32 CInstructionToken::GetTokenType(void){
+  return eToken_instruction;
+}
+

htmlparser/src/nsHTMLTokens.h

@@ -53,7 +53,7 @@ enum eHTMLTokenTypes {
   eToken_unknown=0,
   eToken_start=1,     eToken_end,     eToken_comment,         eToken_entity,
   eToken_whitespace,  eToken_newline, eToken_text,            eToken_attribute,
-  eToken_script,      eToken_style,   eToken_skippedcontent,  
+  eToken_script,      eToken_style,   eToken_skippedcontent,  eToken_instruction,
   eToken_last //make sure this stays the last token...
 };
 
@@ -304,6 +304,23 @@ class CSkippedContentToken: public CAttributeToken {
 };
 
 
+/**
+ *  Whitespace tokens are used where whitespace can be 
+ *  detected as distinct from text. This allows us to 
+ *  easily skip leading/trailing whitespace when desired.
+ *  
+ *  @update  gess 3/25/98
+ */
+class CInstructionToken: public CHTMLToken {
+  public:
+                        CInstructionToken();
+                        CInstructionToken(const nsString& aString);
+    virtual nsresult    Consume(PRUnichar aChar,CScanner& aScanner);
+    virtual const char* GetClassName(void);
+    virtual PRInt32     GetTokenType(void);
+};
+
+
 #endif
 
 

htmlparser/src/nsParserTypes.h

@@ -71,6 +71,7 @@ const PRUint32  kLeftParen        = '(';
 const PRUint32  kRightParen       = ')';
 const PRUint32  kLeftBrace        = '{';
 const PRUint32  kRightBrace       = '}';
+const PRUint32  kQuestionMark     = '?';
 
 
 #endif

parser/htmlparser/src/CNavDTD.cpp

@@ -199,6 +199,7 @@ CToken* nsCTokenRecycler::CreateTokenOfType(eHTMLTokenTypes aType,eHTMLTags aTag
       case eToken_script:     result=new CScriptToken(); break;
       case eToken_style:      result=new CStyleToken(); break;
       case eToken_skippedcontent: result=new CSkippedContentToken(aString); break;
+      case eToken_instruction:result=new CInstructionToken(); break;
         default:
           break;
     }
@@ -308,6 +309,8 @@ PRInt32 NavDispatchTokenHandler(CToken* aToken,nsIDTD* aDTD) {
         result=theDTD->HandleStyleToken(aToken); break;
       case eToken_skippedcontent:
         result=theDTD->HandleSkippedContentToken(aToken); break;
+      case eToken_instruction:
+        result=theDTD->HandleProcessingInstructionToken(aToken); break;
       default:
         result=0;
     }//switch
@@ -337,6 +340,7 @@ void CNavDTD::InitializeDefaultTokenHandlers() {
 //  AddTokenHandler(new CTokenHandler(NavDispatchTokenHandler,eToken_script));
   AddTokenHandler(new CTokenHandler(NavDispatchTokenHandler,eToken_style));
   AddTokenHandler(new CTokenHandler(NavDispatchTokenHandler,eToken_skippedcontent));
+  AddTokenHandler(new CTokenHandler(NavDispatchTokenHandler,eToken_instruction));
 }
 
 
@@ -864,6 +868,21 @@ nsresult CNavDTD::HandleStyleToken(CToken* aToken){
 }
 
 
+/**
+ *  This method gets called when an "instruction" token has been 
+ *  encountered in the parse process. 
+ *  
+ *  @update  gess 3/25/98
+ *  @param   aToken -- next (start) token to be handled
+ *  @return  PR_TRUE if all went well; PR_FALSE if error occured
+ */
+nsresult CNavDTD::HandleProcessingInstructionToken(CToken* aToken){
+  NS_PRECONDITION(0!=aToken,kNullToken);
+
+//  CStyleToken*  st = (CStyleToken*)(aToken);
+  return NS_OK;
+}
+
 /**
  * Retrieve the attributes for this node, and add then into
  * the node.
@@ -2698,9 +2717,15 @@ CNavDTD::ConsumeTag(PRUnichar aChar,CScanner& aScanner,CToken*& aToken) {
           else aToken=gTokenRecycler.CreateTokenOfType(eToken_comment,eHTMLTag_unknown,gEmpty);
         }//if
         break;
+
       case kExclamation:
         aToken=gTokenRecycler.CreateTokenOfType(eToken_comment,eHTMLTag_comment,gEmpty);
         break;
+
+      case kQuestionMark: //it must be an XML processing instruction...
+        aToken=gTokenRecycler.CreateTokenOfType(eToken_instruction,eHTMLTag_unknown,gEmpty);
+        break;
+
       default:
         if(nsString::IsAlpha(aChar))
           return ConsumeStartTag(aChar,aScanner,aToken);

parser/htmlparser/src/CNavDTD.h

@@ -416,6 +416,7 @@ CLASS_EXPORT_HTMLPARS CNavDTD : public nsIDTD {
     nsresult HandleAttributeToken(CToken* aToken);
     nsresult HandleScriptToken(nsCParserNode& aNode);
     nsresult HandleStyleToken(CToken* aToken);
+    nsresult HandleProcessingInstructionToken(CToken* aToken);
 
     virtual  nsITokenRecycler* GetTokenRecycler(void);
 

parser/htmlparser/src/nsHTMLTokens.cpp

@@ -1290,3 +1290,60 @@ const char* GetTagName(PRInt32 aTag) {
   }
   return result;
 }
+
+
+/**
+ *  
+ *  
+ *  @update  gess 9/23/98
+ *  @param   
+ *  @return  
+ */
+CInstructionToken::CInstructionToken() : CHTMLToken(eHTMLTag_unknown) {
+}
+
+/**
+ *  
+ *  
+ *  @update  gess 9/23/98
+ *  @param   
+ *  @return  
+ */
+CInstructionToken::CInstructionToken(const nsString& aString) : CHTMLToken(aString) {
+}
+
+/**
+ *  
+ *  
+ *  @update  gess 9/23/98
+ *  @param   
+ *  @return  
+ */
+nsresult CInstructionToken::Consume(PRUnichar aChar,CScanner& aScanner){
+  mTextValue="<?";
+  nsresult result=aScanner.ReadUntil(mTextValue,kGreaterThan,PR_TRUE);
+  return result;
+}
+
+/**
+ *  
+ *  
+ *  @update  gess 9/23/98
+ *  @param   
+ *  @return  
+ */
+const char* CInstructionToken::GetClassName(void){
+  return "instruction";
+}
+
+/**
+ *  
+ *  
+ *  @update  gess 9/23/98
+ *  @param   
+ *  @return  
+ */
+PRInt32 CInstructionToken::GetTokenType(void){
+  return eToken_instruction;
+}
+

parser/htmlparser/src/nsHTMLTokens.h

@@ -53,7 +53,7 @@ enum eHTMLTokenTypes {
   eToken_unknown=0,
   eToken_start=1,     eToken_end,     eToken_comment,         eToken_entity,
   eToken_whitespace,  eToken_newline, eToken_text,            eToken_attribute,
-  eToken_script,      eToken_style,   eToken_skippedcontent,  
+  eToken_script,      eToken_style,   eToken_skippedcontent,  eToken_instruction,
   eToken_last //make sure this stays the last token...
 };
 
@@ -304,6 +304,23 @@ class CSkippedContentToken: public CAttributeToken {
 };
 
 
+/**
+ *  Whitespace tokens are used where whitespace can be 
+ *  detected as distinct from text. This allows us to 
+ *  easily skip leading/trailing whitespace when desired.
+ *  
+ *  @update  gess 3/25/98
+ */
+class CInstructionToken: public CHTMLToken {
+  public:
+                        CInstructionToken();
+                        CInstructionToken(const nsString& aString);
+    virtual nsresult    Consume(PRUnichar aChar,CScanner& aScanner);
+    virtual const char* GetClassName(void);
+    virtual PRInt32     GetTokenType(void);
+};
+
+
 #endif
 
 

parser/htmlparser/src/nsParserTypes.h

@@ -71,6 +71,7 @@ const PRUint32  kLeftParen        = '(';
 const PRUint32  kRightParen       = ')';
 const PRUint32  kLeftBrace        = '{';
 const PRUint32  kRightBrace       = '}';
+const PRUint32  kQuestionMark     = '?';
 
 
 #endif