Add attribute name validation to the validator

Author: kata198

AdvancedHTMLParser/Validator.py

@@ -2,10 +2,13 @@
 # Copyright (c) 2015, 2019 Tim Savannah All Rights Rserved under LGPLv3. See LICENSE (https://gnu.org/licenses/lgpl-3.0.txt) for more information.
 
 from .Parser import AdvancedHTMLParser
+from .Tags import isValidAttributeName
 
-from .exceptions import InvalidCloseException, MissedCloseException
+from .exceptions import InvalidCloseException, MissedCloseException, InvalidAttributeNameException
 
-__all__ = ('InvalidCloseException', 'MissedCloseException', 'ValidatingAdvancedHTMLParser')
+__all__ = ('InvalidCloseException', 'MissedCloseException', 'InvalidAttributeNameException',
+    'ValidatingAdvancedHTMLParser',
+)
 
 class ValidatingAdvancedHTMLParser(AdvancedHTMLParser):
     '''
@@ -16,6 +19,27 @@ class ValidatingAdvancedHTMLParser(AdvancedHTMLParser):
         exceptions.MissedCloseException  - The parsed string/file missed closing an item.
     '''
 
+    def handle_starttag(self, tagName, attributeList, isSelfClosing=False):
+        '''
+            handle_starttag - internal for parsing,
+
+                ValidatingAdvancedHTMLParser will run through the attributes list and make sure
+                  none have an invalid name, or will raise an error.
+
+
+                @raises - InvalidAttributeNameException if an attribute name is passed with invalid character(s)
+        '''
+
+        # Iterate over the passed attributes, and validate them.
+        for (attrName, attrValue) in attributeList:
+
+            if isValidAttributeName(attrName) is False:
+
+                raise InvalidAttributeNameException(tagName, attrName, attrValue)
+
+        # Done validating, feed to parent.
+        return AdvancedHTMLParser.handle_starttag(self, tagName, attributeList, isSelfClosing)
+
 
     def handle_endtag(self, tagName):
         '''

AdvancedHTMLParser/exceptions.py

@@ -32,7 +32,7 @@ class HTMLValidationException(Exception):
 
 class InvalidCloseException(HTMLValidationException):
     '''
-        InvalidCloseException - Raised when a tag is closed that shouldn't be closed.
+        InvalidCloseException - Raised when a tag is closed that shouldn't be closed in validating parser
     '''
 
     def __init__(self, triedToClose, stillOpen):
@@ -48,7 +48,7 @@ def __init__(self, triedToClose, stillOpen):
 
 class MissedCloseException(HTMLValidationException):
     '''
-        MissedCloseException - Raised when a close was missed
+        MissedCloseException - Raised when a close was missed in validating parser
     '''
 
     def __init__(self, triedToClose, stillOpen):
@@ -60,6 +60,29 @@ def __init__(self, triedToClose, stillOpen):
         Exception.__init__(self, message)
 
 
+class InvalidAttributeNameException(HTMLValidationException):
+    '''
+        InvalidAttributeNameException - Raised when an invalid attribute name is found when parsing via validating parser
+    '''
+
+    def __init__(self, tagName, badAttributeName, badAttributeValue):
+        '''
+            __init__ - Create this object
+
+                @param tagName <str> - Tag name
+
+                @param badAttributeName <str> - Bad attribute name
+
+                @param badAttributeValue <str> - Bad attribute value
+        '''
+
+        message = 'Parsed a tag %s  which contains an invalid attribute, %s = %s . ( Maybe characters outside quotes in tag? )' % ( \
+            tagName, repr(badAttributeName), repr(badAttributeValue) \
+        )
+
+        Exception.__init__(self, message)
+
+
 class IndexSizeErrorException(ValueError):
 
     def __init__(self, *args, **kwargs):