@@ -71,6 +71,8 @@ def __init__(self, stream, encoding=None, parseMeta=True, useChardet=True,
7171 "comment" :self .commentState ,
7272 "commentEndDash" :self .commentEndDashState ,
7373 "commentEnd" :self .commentEndState ,
74+ "commentEndBang" :self .commentEndBangState ,
75+ "commentEndSpace" :self .commentEndSpaceState ,
7476 "doctype" :self .doctypeState ,
7577 "beforeDoctypeName" :self .beforeDoctypeNameState ,
7678 "doctypeName" :self .doctypeNameState ,
@@ -476,7 +478,7 @@ def tagNameState(self):
476478 elif data is EOF :
477479 self .tokenQueue .append ({"type" : tokenTypes ["ParseError" ], "data" :
478480 "eof-in-tag-name" })
479- self .emitCurrentToken ()
481+ self .state = self . states [ "data" ]
480482 elif data == u"/" :
481483 self .state = self .states ["selfClosingStartTag" ]
482484 else :
@@ -504,7 +506,7 @@ def beforeAttributeNameState(self):
504506 elif data is EOF :
505507 self .tokenQueue .append ({"type" : tokenTypes ["ParseError" ], "data" :
506508 "expected-attribute-name-but-got-eof" })
507- self .emitCurrentToken ()
509+ self .state = self . states [ "data" ]
508510 else :
509511 self .currentToken ["data" ].append ([data , "" ])
510512 self .state = self .states ["attributeName" ]
@@ -654,7 +656,7 @@ def attributeValueUnQuotedState(self):
654656 self .processEntityInAttribute (None )
655657 elif data == u">" :
656658 self .emitCurrentToken ()
657- elif data == u'"' or data == u"'" or data == u"=" :
659+ elif data in ( u'"' , u"'" , u"=" , u"<" ) :
658660 self .tokenQueue .append ({"type" : tokenTypes ["ParseError" ], "data" :
659661 "unexpected-character-in-unquoted-attribute-value" })
660662 self .currentToken ["data" ][- 1 ][1 ] += data
@@ -697,7 +699,6 @@ def selfClosingStartTagState(self):
697699 self .tokenQueue .append ({"type" : tokenTypes ["ParseError" ],
698700 "data" :
699701 "unexpected-EOF-after-solidus-in-tag" })
700- self .emitCurrentToken ()
701702 self .stream .unget (data )
702703 self .state = self .states ["data" ]
703704 else :
@@ -846,6 +847,15 @@ def commentEndState(self):
846847 self .tokenQueue .append ({"type" : tokenTypes ["ParseError" ], "data" :
847848 "unexpected-dash-after-double-dash-in-comment" })
848849 self .currentToken ["data" ] += data
850+ elif data in spaceCharacters :
851+ self .currentToken ["data" ] += "--" + data
852+ self .tokenQueue .append ({"type" : tokenTypes ["ParseError" ], "data" :
853+ "unexpected-space-after-double-dash-in-comment" })
854+ self .state = self .states ["commentEndSpace" ]
855+ elif data == "!" :
856+ self .tokenQueue .append ({"type" : tokenTypes ["ParseError" ], "data" :
857+ "unexpected-bang-after-double-dash-in-comment" })
858+ self .state = self .states ["commentEndBang" ]
849859 elif data is EOF :
850860 self .tokenQueue .append ({"type" : tokenTypes ["ParseError" ], "data" :
851861 "eof-in-comment-double-dash" })
@@ -859,10 +869,53 @@ def commentEndState(self):
859869 self .state = self .states ["comment" ]
860870 return True
861871
872+ def commentEndBangState (self ):
873+ data = self .stream .char ()
874+ if data == u">" :
875+ self .tokenQueue .append (self .currentToken )
876+ self .state = self .states ["data" ]
877+ elif data == u"-" :
878+ self .currentToken ["data" ] += "--!"
879+ self .state = self .states ["commentEndDash" ]
880+ elif data is EOF :
881+ self .tokenQueue .append ({"type" : tokenTypes ["ParseError" ], "data" :
882+ "eof-in-comment-end-bang-state" })
883+ self .tokenQueue .append (self .currentToken )
884+ self .state = self .states ["data" ]
885+ else :
886+ self .currentToken ["data" ] += u"--!" + data
887+ self .state = self .states ["comment" ]
888+ return True
889+
890+ def commentEndSpaceState (self ):
891+ data = self .stream .char ()
892+ if data == u">" :
893+ self .tokenQueue .append (self .currentToken )
894+ self .state = self .states ["data" ]
895+ elif data == u"-" :
896+ self .state = self .states ["commentEndDash" ]
897+ elif data in spaceCharacters :
898+ self .currentToken ["data" ] += data
899+ elif data is EOF :
900+ self .tokenQueue .append ({"type" : tokenTypes ["ParseError" ], "data" :
901+ "eof-in-comment-end-space-state" })
902+ self .tokenQueue .append (self .currentToken )
903+ self .state = self .states ["data" ]
904+ else :
905+ self .currentToken ["data" ] += data
906+ self .state = self .states ["comment" ]
907+ return True
908+
862909 def doctypeState (self ):
863910 data = self .stream .char ()
864911 if data in spaceCharacters :
865912 self .state = self .states ["beforeDoctypeName" ]
913+ elif data is EOF :
914+ self .tokenQueue .append ({"type" : tokenTypes ["ParseError" ], "data" :
915+ "expected-doctype-name-but-got-eof" })
916+ self .currentToken ["correct" ] = False
917+ self .tokenQueue .append (self .currentToken )
918+ self .state = self .states ["data" ]
866919 else :
867920 self .tokenQueue .append ({"type" : tokenTypes ["ParseError" ], "data" :
868921 "need-space-after-doctype" })
0 commit comments